matching.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Record Linkage"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![img](./media/linkage10.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![img](./media/linkage1.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![img](./media/linkage2.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![img](./media/linkage3.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![img](./media/linkage4.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![img](./media/linkage5.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![img](./media/linkage6.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![img](./media/linkage7.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![img](./media/linkage8.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![img](./media/linkage9.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![img](./media/linkage11.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {},
   "outputs": [],
   "source": [
    "import csv \n",
    "import json\n",
    "import string\n",
    "import random\n",
    "import typing\n",
    "\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import spacy\n",
    "from sklearn.model_selection import train_test_split\n",
    "import keras\n",
    "import seaborn as sns\n",
    "from sklearn.metrics import classification_report, confusion_matrix, roc_curve, auc, roc_auc_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# python3 -m spacy download en_core_web_sm\n",
    "english: spacy.lang.en.English = spacy.load(\"en_core_web_sm\")\n",
    "    \n",
    "np.random.seed(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_string(size: int = 10) -> str:\n",
    "    \"\"\"\n",
    "    Generates a random string.\n",
    "    \"\"\"\n",
    "    chars: str = string.ascii_lowercase + \"aeiou\" * 30 + \"rtsmpcnlgr\" * 15\n",
    "    return \"\".join([\n",
    "        random.choice(chars)\n",
    "        for _ in range(size)\n",
    "    ])\n",
    "\n",
    "assert isinstance(random_string(), str)\n",
    "assert random_string() != random_string()\n",
    "assert len(random_string(10)) == 10\n",
    "assert len(random_string(20)) == 20"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_integer(low: int = 0, high: int = 100) -> int:\n",
    "    \"\"\"\n",
    "    Generates a random integer.\n",
    "    \"\"\"\n",
    "    return random.randint(low, high)\n",
    "\n",
    "assert isinstance(random_integer(), int)\n",
    "assert random_integer() >= 0\n",
    "assert random_integer() <= 100\n",
    "assert random_integer(low=120, high=140) >= 120\n",
    "assert random_integer(low=120, high=140) <= 140"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_float(low: int = 0, high: int = 100) -> float:\n",
    "    \"\"\"\n",
    "    Generates a random float number.\n",
    "    \"\"\"\n",
    "    return random.uniform(low, high)\n",
    "\n",
    "assert isinstance(random_float(), float)\n",
    "assert random_float() >= 0\n",
    "assert random_float() <= 100\n",
    "assert random_float(low=120, high=140) <= 140\n",
    "assert random_float(low=120, high=140) >= 120"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_probability() -> float:\n",
    "    \"\"\"\n",
    "    Generates a random probability float number.\n",
    "    \"\"\"\n",
    "    return random.uniform(0, 1)\n",
    "\n",
    "assert isinstance(random_probability(), float)\n",
    "assert random_probability() >= 0\n",
    "assert random_probability() <= 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_event(probability: float = 0.5) -> bool:\n",
    "    \"\"\"\n",
    "    Generates a determines the occurrence of a random event.\n",
    "    \"\"\"\n",
    "    assert probability >= 0\n",
    "    assert probability <= 1\n",
    "    return random_probability() < probability\n",
    "\n",
    "assert isinstance(random_event(), bool)\n",
    "assert random_event(probability=1) is True\n",
    "assert random_event(probability=0) is False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_option(options: list) -> str:\n",
    "    \"\"\"\n",
    "    Selects a value from a list randomly.\n",
    "    \"\"\"\n",
    "    return random.choice(list(set(options)))\n",
    "\n",
    "assert random_option([1, 2, 3]) in [1, 2, 3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_phone() -> str:\n",
    "    \"\"\"\n",
    "    Generates a random phone number.\n",
    "    \"\"\"\n",
    "    return \"-\".join(\n",
    "        [\n",
    "            str(random_integer(100, 999)),\n",
    "            str(random_integer(100, 999)),\n",
    "            str(random_integer(1000, 9999)),\n",
    "        ]\n",
    "    )\n",
    "\n",
    "assert isinstance(random_phone(), str)\n",
    "assert random_phone() != random_phone()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_name() -> str:\n",
    "    \"\"\"\n",
    "    Generates a random name.\n",
    "    \"\"\"\n",
    "    return \" \".join(\n",
    "        [\n",
    "            random_string(random_integer(8, 12)),\n",
    "            random_string(random_integer(8, 12)),\n",
    "        ]\n",
    "    ).title()\n",
    "\n",
    "assert isinstance(random_name(), str)\n",
    "assert random_name() != random_name()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_city() -> str:\n",
    "    \"\"\"\n",
    "    Generates a random city name.\n",
    "    \"\"\"\n",
    "    return \"\".join(\n",
    "        [\n",
    "            random_string(random_integer(3, 6)),\n",
    "            random_option([\" \", \"\"]),\n",
    "            random_string(random_integer(3, 6)),\n",
    "        ]\n",
    "    ).title()\n",
    "\n",
    "assert isinstance(random_city(), str)\n",
    "assert random_city() != random_city()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_website() -> str:\n",
    "    \"\"\"\n",
    "    Generates a random website.\n",
    "    \"\"\"\n",
    "    return \".\".join(\n",
    "        [\n",
    "            \"www\",\n",
    "            random_string(random_integer(7, 15)),\n",
    "            random_option([\"com\", \"net\", \"org\"]),\n",
    "        ]\n",
    "    ).lower()\n",
    "\n",
    "assert isinstance(random_website(), str)\n",
    "assert random_website() != random_website()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_linkedin_url() -> str:\n",
    "    \"\"\"\n",
    "    Generates a random LinkedIn URL.\n",
    "    \"\"\"\n",
    "    return \"/\".join(\n",
    "        [\n",
    "            \"www.linkedin.com\",\n",
    "            \"u\",\n",
    "            random_string(random_integer(8, 15)),\n",
    "        ]\n",
    "    ).lower()\n",
    "\n",
    "assert isinstance(random_linkedin_url(), str)\n",
    "assert random_linkedin_url() != random_linkedin_url()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_facebook_url() -> str:\n",
    "    \"\"\"\n",
    "    Generates a random Facebook URL.\n",
    "    \"\"\"\n",
    "    return \"/\".join(\n",
    "        [\n",
    "            \"https://\",\n",
    "            \"www.facebook.com\",\n",
    "            \"profile\",\n",
    "            random_string(random_integer(8, 15)),\n",
    "        ]\n",
    "    ).lower()\n",
    "\n",
    "assert isinstance(random_facebook_url(), str)\n",
    "assert random_facebook_url() != random_facebook_url()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_country() -> str:\n",
    "    \"\"\"\n",
    "    Generates a random country.\n",
    "    \"\"\"\n",
    "    return \"\".join(\n",
    "        [\n",
    "            random_string(random_integer(3, 6)),\n",
    "            random_option([\" \", \"\"]),\n",
    "            random_string(random_integer(3, 6)),\n",
    "        ]\n",
    "    ).title()\n",
    "\n",
    "assert isinstance(random_country(), str)\n",
    "assert random_country() != random_country()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_street() -> str:\n",
    "    \"\"\"\n",
    "    Generates a random street address.\n",
    "    \"\"\"\n",
    "    return \" \".join(\n",
    "        [\n",
    "            str(random_integer(1, 999)),\n",
    "            random_string(random_integer(3, 8)),\n",
    "            random_string(random_integer(4, 6)) + \",\",\n",
    "            random_string(random_integer(4, 6)),\n",
    "            random_option([\"St.\", \"Av.\"]),\n",
    "        ]\n",
    "    ).title()\n",
    "\n",
    "assert isinstance(random_street(), str)\n",
    "assert random_street() != random_street()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "def similar_string(value: str, mutability: float = 0.2) -> str:\n",
    "    \"\"\"\n",
    "    Generates a random string similar to a given string.\n",
    "    \"\"\"\n",
    "    new_value: list = list(value)\n",
    "    for i in range(len(new_value)):\n",
    "        if random_event(mutability):\n",
    "            new_value[i] = random_string(1)\n",
    "    if new_value and random_event(mutability):\n",
    "        new_value.pop(-1)\n",
    "    if new_value and random_event(mutability):\n",
    "        new_value.append(random_string(1))\n",
    "    if not new_value and random_event(mutability):\n",
    "        new_value: str = random_string(random_integer(4, 12))\n",
    "    return \"\".join(new_value)\n",
    "\n",
    "assert isinstance(similar_string(\"a\", mutability=0.5), str)\n",
    "assert similar_string(\"a\", mutability=0.0) == \"a\"\n",
    "assert similar_string(\"a\", mutability=1.0) != \"a\"\n",
    "assert similar_string(\"aaaaaa\", mutability=1.0) != similar_string(\"aaaaaa\", mutability=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "def similar_dict(record: dict, mutability: float = 0.1) -> dict:\n",
    "    \"\"\"\n",
    "    Generates a random dictionary similar to a given dictionary.\n",
    "    \"\"\"\n",
    "    other: dict = {}\n",
    "    for k, v in record.items():\n",
    "        if isinstance(v, str):\n",
    "            other[k] = similar_string(value=v, mutability=mutability)\n",
    "        else:\n",
    "            other[k] = v\n",
    "    return other\n",
    "\n",
    "assert isinstance(similar_dict({\"a\": \"b\"}, mutability=0.5), dict)\n",
    "assert similar_dict({\"a\": \"b\"}, mutability=0.0) == {\"a\": \"b\"}\n",
    "assert similar_dict({\"a\": \"b\"}, mutability=1.0) != {\"a\": \"b\"}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_company(nulls: float = 0.1) -> dict:\n",
    "    \"\"\"\n",
    "    Generates a random dictionary representation of a company.\n",
    "    \"\"\"\n",
    "    return {\n",
    "        \"city\": \"\" if random_event(probability=nulls) else random_city(),\n",
    "        \"country\": \"\" if random_event(probability=nulls) else random_country(),\n",
    "        \"company_name\": random_name(),\n",
    "        \"phone\": \"\" if random_event(probability=nulls) else random_phone(),\n",
    "        \"street_address\": \"\" if random_event(probability=nulls) else random_street(),\n",
    "        \"website\": \"\" if random_event(probability=nulls) else random_website(),\n",
    "        \"linkedin_url\": \"\" if random_event(probability=nulls) else random_linkedin_url(),\n",
    "        \"facebook_url\": \"\" if random_event(probability=nulls) else random_facebook_url(),\n",
    "    }\n",
    "\n",
    "assert isinstance(random_company(), dict)\n",
    "assert all(isinstance(k, str) for k in random_company().keys())\n",
    "assert all(isinstance(v, str) for v in random_company().values())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def random_person(nulls: float = 0.1) -> dict:\n",
    "    \"\"\"\n",
    "    Generates a random representation of a person.\n",
    "    \"\"\"\n",
    "    return {\n",
    "        \"city\": \"\" if random_event(probability=nulls) else random_city(),\n",
    "        \"country\": \"\" if random_event(probability=nulls) else random_country(),\n",
    "        \"name\": random_name(),\n",
    "        \"phone\": \"\" if random_event(probability=nulls) else random_phone(),\n",
    "        \"street_address\": \"\" if random_event(probability=nulls) else random_street(),\n",
    "        \"linkedin_url\": \"\" if random_event(probability=nulls) else random_linkedin_url(),\n",
    "        \"facebook_url\": \"\" if random_event(probability=nulls) else random_facebook_url(),\n",
    "    }\n",
    "\n",
    "assert isinstance(random_person(), dict)\n",
    "assert all(isinstance(k, str) for k in random_person().keys())\n",
    "assert all(isinstance(v, str) for v in random_person().values())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def add_suffix(record: dict, suffix: str) -> dict:\n",
    "    \"\"\"\n",
    "    Appends a suffix to all the keys of a given dictionary.\n",
    "    \"\"\"\n",
    "    return {f\"{k}#{suffix}\": v for k, v in record.items()}\n",
    "\n",
    "assert add_suffix({\"a\": \"b\"}, \"1\") == {\"a#1\": \"b\"}\n",
    "assert add_suffix({\"a\": \"b\"}, \"2\") == {\"a#2\": \"b\"}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def word2vec(text: typing.Optional[str]) -> np.ndarray:\n",
    "    \"\"\"\n",
    "    Transforms a string into a numpy array of a fixed size.\n",
    "    \"\"\"\n",
    "    if not isinstance(text, str) or text == \"\":\n",
    "        return np.zeros(96)\n",
    "    return english(text).vector\n",
    "\n",
    "assert isinstance(word2vec(\"lorem\"), np.ndarray)\n",
    "assert np.array_equal(word2vec(\"lorem\"), word2vec(\"lorem\"))\n",
    "assert word2vec(\"lorem\").shape == (96, )\n",
    "assert isinstance(word2vec(\"\"), np.ndarray)\n",
    "assert np.array_equal(word2vec(\"\"), word2vec(\"\"))\n",
    "assert word2vec(\"\").shape == (96, )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Vectorizing: a\n",
      "Vectorizing: t\n",
      "Vectorizing: a\n",
      "Vectorizing: t\n",
      "Vectorizing: a\n",
      "Vectorizing: t\n"
     ]
    }
   ],
   "source": [
    "def df2vec(dataset: pd.DataFrame, target: str = 'match') -> pd.DataFrame:\n",
    "    \"\"\"\n",
    "    Transforms a string into a numpy array of a fixed size.\n",
    "    \"\"\"\n",
    "    vectorized_dataset: pd.DataFrame = dataset.copy()  \n",
    "    for feature in vectorized_dataset.columns:\n",
    "        print(\"Vectorizing:\", feature)\n",
    "        if feature != target:\n",
    "            vectorized_dataset[feature] = vectorized_dataset[feature].apply(word2vec).apply(np.array)\n",
    "    return vectorized_dataset\n",
    "\n",
    "assert isinstance(df2vec(\n",
    "    pd.DataFrame({\"a\": [\"b\", \"c\"], \"t\": [0, 1]}), target=\"t\",\n",
    ")['a'][0], np.ndarray)\n",
    "assert isinstance(df2vec(\n",
    "    pd.DataFrame({\"a\": [\"b\", \"c\"], \"t\": [0, 1]}), target=\"t\",\n",
    ")['a'][1], np.ndarray)\n",
    "assert df2vec(\n",
    "    pd.DataFrame({\"a\": [\"b\", \"c\"], \"t\": [0, 1]}), target=\"t\",\n",
    ").shape == (2, 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Companies: (1, 17)\n",
      "Companies: (1, 17)\n",
      "Companies: (1, 17)\n",
      "Companies: (1, 17)\n",
      "Companies: (2, 17)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def generate_random_companies(\n",
    "    total: int = 1000,\n",
    "    target: str = 'match',\n",
    "    matches: float = 0.5,\n",
    "    nulls: float = 0.1,\n",
    "    mutability: float = 0.01,\n",
    "    feedback: list = None,\n",
    "    relevance: int = 10,\n",
    ") -> pd.DataFrame:\n",
    "    \"\"\"\n",
    "    Generates a random dataset of $total companies.\n",
    "    \"\"\"\n",
    "    raw_companies_list: typing.List[dict] = []\n",
    "    for _ in range(total):\n",
    "        company1: dict = random_company(nulls=nulls)\n",
    "        if random_event(probability=matches):\n",
    "            company2: dict = similar_dict(company1, mutability=mutability)\n",
    "            match: str = \"1\" \n",
    "        else:\n",
    "            company2: dict = random_company(nulls=0.1)\n",
    "            match: str = \"0\" \n",
    "        row: dict = {}\n",
    "        for k, v in add_suffix(company1, \"1\").items():\n",
    "            row[k] = v \n",
    "        for k, v in add_suffix(company2, \"2\").items():\n",
    "            row[k] = v \n",
    "        raw_companies_list.append({\n",
    "            **add_suffix(company1, \"1\"),\n",
    "            **add_suffix(company2, \"2\"),\n",
    "            target: match,\n",
    "        })\n",
    "    if feedback is not None:\n",
    "        for case in feedback:\n",
    "            raw_companies_list.extend([{\n",
    "                **add_suffix(case[0], \"1\"),\n",
    "                **add_suffix(case[1], \"2\"),\n",
    "                target: case[2],                \n",
    "            }] * relevance)\n",
    "    raw_companies_dataframe: pd.DataFrame = pd.DataFrame(raw_companies_list)\n",
    "    raw_companies_dataframe[target] = raw_companies_dataframe[target].astype(int)\n",
    "    print(\"Companies:\", raw_companies_dataframe.shape)\n",
    "    raw_companies_dataframe[target].astype(int).value_counts().plot(kind='pie')\n",
    "    return raw_companies_dataframe\n",
    "\n",
    "assert isinstance(generate_random_companies(total=1), pd.DataFrame)\n",
    "assert not generate_random_companies(total=1).equals(generate_random_companies(total=1))\n",
    "assert generate_random_companies(total=1).shape == (1, 17)\n",
    "assert generate_random_companies(total=2).shape == (2, 17)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Persons: (1, 15)\n",
      "Persons: (1, 15)\n",
      "Persons: (1, 15)\n",
      "Persons: (1, 15)\n",
      "Persons: (2, 15)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def generate_random_persons(\n",
    "    total: int = 1000,\n",
    "    target: str = 'match',\n",
    "    matches: float = 0.5,\n",
    "    nulls: float = 0.1,\n",
    "    mutability: float = 0.01,\n",
    "    feedback: list = None,\n",
    "    relevance: int = 10,\n",
    ") -> pd.DataFrame:\n",
    "    \"\"\"\n",
    "    Generates a random dataset of $total persons.\n",
    "    \"\"\"\n",
    "    raw_persons_list: typing.List[dict] = []\n",
    "    for _ in range(total):\n",
    "        person1: dict = random_person(nulls=nulls)\n",
    "        if random_event(probability=matches):\n",
    "            person2: dict = similar_dict(person1, mutability=mutability)\n",
    "            match: str = \"1\" \n",
    "        else:\n",
    "            person2: dict = random_person(nulls=0.1)\n",
    "            match: str = \"0\" \n",
    "        row: dict = {}\n",
    "        for k, v in add_suffix(person1, \"1\").items():\n",
    "            row[k] = v \n",
    "        for k, v in add_suffix(person2, \"2\").items():\n",
    "            row[k] = v \n",
    "        raw_persons_list.append({\n",
    "            **add_suffix(person1, \"1\"),\n",
    "            **add_suffix(person2, \"2\"),\n",
    "            target: match,\n",
    "        })\n",
    "    if feedback is not None:\n",
    "        for case in feedback:\n",
    "            raw_persons_list.extend([{\n",
    "                **add_suffix(case[0], \"1\"),\n",
    "                **add_suffix(case[1], \"2\"),\n",
    "                target: case[2],                \n",
    "            }] * relevance)\n",
    "    raw_persons_dataframe: pd.DataFrame = pd.DataFrame(raw_persons_list)\n",
    "    raw_persons_dataframe[target] = raw_persons_dataframe[target].astype(int)\n",
    "    print(\"Persons:\", raw_persons_dataframe.shape)\n",
    "    raw_persons_dataframe[target].astype(int).value_counts().plot(kind='pie')\n",
    "    return raw_persons_dataframe\n",
    "\n",
    "assert isinstance(generate_random_persons(total=1), pd.DataFrame)\n",
    "assert not generate_random_persons(total=1).equals(generate_random_persons(total=1))\n",
    "assert generate_random_persons(total=1).shape == (1, 15)\n",
    "assert generate_random_persons(total=2).shape == (2, 15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "def vec2matrix(data: pd.DataFrame) -> np.ndarray:\n",
    "    \"\"\"\n",
    "    Transforms a Pandas DataFrame into a Numpy 3D Array.\n",
    "    \"\"\"\n",
    "    features: typing.List[np.ndarray] = []\n",
    "    for feature in data.columns:\n",
    "        features.append(np.array(data[feature].to_list()))\n",
    "    matrix: np.array = np.array(features)\n",
    "    matrix: np.array = np.swapaxes(matrix, 0, 1)\n",
    "    return matrix\n",
    "\n",
    "array: np.ndarray = np.zeros((3, ))\n",
    "dataframe: pd.DataFrame = pd.DataFrame({\"a\": [array, array]})\n",
    "assert vec2matrix(dataframe).shape == (2, 1, 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predictors: ['a']\n",
      "Predicted: m\n",
      "X (20, 1)\n",
      "Y (20,)\n",
      "Train (12, 1) (12,)\n",
      "Test (4, 1) (4,)\n",
      "Val (4, 1) (4,)\n",
      "Train: (12, 1, 3) (12,)\n",
      "Test: (4, 1, 3) (4,)\n",
      "Val: (4, 1, 3) (4,)\n",
      "Matches: [6, 2, 2] [12, 4, 4]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def split_dataframe(\n",
    "    dataset: pd.DataFrame,\n",
    "    target: str = 'match',\n",
    "    test_size: float = 0.2,\n",
    "    val_size: float = 0.2,\n",
    "    random_state: int = 5,\n",
    ") -> typing.Tuple[np.ndarray]:\n",
    "    \"\"\"\n",
    "    Splits a dataset into Train, Test and Validation.\n",
    "    \"\"\"\n",
    "\n",
    "    features: pd.Index = [\n",
    "        column\n",
    "        for column in dataset.columns\n",
    "        if column != target\n",
    "    ]\n",
    "\n",
    "    print(\"Predictors:\", features)\n",
    "    print(\"Predicted:\", target)\n",
    "    \n",
    "    x: pd.DataFrame = dataset[features]\n",
    "    y: pd.DataFrame = dataset[target]\n",
    "    assert x.shape[0] == y.shape[0]\n",
    "        \n",
    "    print(\"X\", x.shape)\n",
    "    print(\"Y\", y.shape)\n",
    "    \n",
    "    x_train, x_test, y_train, y_test = train_test_split(\n",
    "        x,\n",
    "        y,\n",
    "        test_size=test_size,\n",
    "        random_state=random_state,\n",
    "        shuffle=True,\n",
    "        stratify=y,\n",
    "    )\n",
    "    x_train, x_val, y_train, y_val = train_test_split(\n",
    "        x_train,\n",
    "        y_train,\n",
    "        test_size=val_size,\n",
    "        random_state=random_state,\n",
    "        shuffle=True,\n",
    "        stratify=y_train,\n",
    "    )\n",
    "    \n",
    "    print(\"Train\", x_train.shape, y_train.shape)\n",
    "    print(\"Test\", x_test.shape, y_test.shape)\n",
    "    print(\"Val\", x_val.shape, y_val.shape)\n",
    "\n",
    "    assert x_train.shape[0] == y_train.shape[0]\n",
    "    assert x_test.shape[0] == y_test.shape[0]\n",
    "    assert x_val.shape[0] == y_val.shape[0]\n",
    "\n",
    "    assert x_train.shape[1] == x_test.shape[1]\n",
    "    assert x_train.shape[1] == x_val.shape[1]\n",
    "    assert x_test.shape[1] == x_val.shape[1]\n",
    "\n",
    "    assert x_train.shape[0] + x_test.shape[0] + x_val.shape[0] == x.shape[0]\n",
    "    assert y_train.shape[0] + y_test.shape[0] + y_val.shape[0] == y.shape[0]\n",
    "\n",
    "    y_train: np.ndarray = y_train.values\n",
    "    y_test: np.ndarray = y_test.values\n",
    "    y_val: np.ndarray = y_val.values\n",
    "    \n",
    "    x_train: typing.List[np.array] = vec2matrix(x_train)\n",
    "    x_test: typing.List[np.array] = vec2matrix(x_test)\n",
    "    x_val: typing.List[np.array] = vec2matrix(x_val)\n",
    "\n",
    "    print(\"Train:\", x_train.shape, y_train.shape)\n",
    "    print(\"Test:\", x_test.shape, y_test.shape)\n",
    "    print(\"Val:\", x_val.shape, y_val.shape)\n",
    "        \n",
    "    matches: list = [\n",
    "        sum(y_train == 1),\n",
    "        sum(y_test == 1),\n",
    "        sum(y_val == 1),   \n",
    "    ]\n",
    "    missmatches: list = [\n",
    "        y_train.shape[0],\n",
    "        y_test.shape[0],\n",
    "        y_val.shape[0],   \n",
    "    ]\n",
    "    print(\"Matches:\", matches, missmatches)\n",
    "    assert matches != missmatches, \"Too many matches!\"\n",
    "    assert all(matches), \"Not enough matches!\"\n",
    "    assert all(missmatches), \"Not enough missmatches!\"\n",
    "    labels: list = ['Train', 'Test', 'Val']\n",
    "    fig, ax = plt.subplots()\n",
    "    ax.bar(labels, missmatches, label='0')\n",
    "    ax.bar(labels, matches, label='1')\n",
    "    ax.set_ylabel('Records')\n",
    "    ax.set_title('Datasets')\n",
    "    ax.legend()\n",
    "    plt.show()\n",
    "    \n",
    "    return x_train, y_train, x_test, y_test, x_val, y_val\n",
    "\n",
    "\n",
    "x_train, y_train, x_test, y_test, x_val, y_val = split_dataframe(\n",
    "    dataset=pd.DataFrame({\"a\": [np.array([1, 2, 3])] * 20, \"m\": [1, 0] * 10}),\n",
    "    target='m',\n",
    "    test_size=0.2,\n",
    "    val_size=0.2,\n",
    ")\n",
    "assert x_train.shape == (12, 1, 3)\n",
    "assert y_train.shape == (12, )\n",
    "assert x_test.shape == (4, 1, 3)\n",
    "assert y_test.shape == (4, )\n",
    "assert x_val.shape == (4, 1, 3)\n",
    "assert y_val.shape == (4, )\n",
    "assert isinstance(x_train, np.ndarray)\n",
    "assert isinstance(y_train, np.ndarray)\n",
    "assert isinstance(x_test, np.ndarray)\n",
    "assert isinstance(y_test, np.ndarray)\n",
    "assert isinstance(x_val, np.ndarray)\n",
    "assert isinstance(y_val, np.ndarray)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input: (2, 2, 2)\n",
      "Model: \"model_12\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_13 (InputLayer)        [(None, 2, 2)]            0         \n",
      "_________________________________________________________________\n",
      "flatten_12 (Flatten)         (None, 4)                 0         \n",
      "_________________________________________________________________\n",
      "dense_60 (Dense)             (None, 4096)              20480     \n",
      "_________________________________________________________________\n",
      "dropout_48 (Dropout)         (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dense_61 (Dense)             (None, 2048)              8390656   \n",
      "_________________________________________________________________\n",
      "dropout_49 (Dropout)         (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_62 (Dense)             (None, 1024)              2098176   \n",
      "_________________________________________________________________\n",
      "dropout_51 (Dropout)         (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_64 (Dense)             (None, 1)                 1025      \n",
      "=================================================================\n",
      "Total params: 10,510,337\n",
      "Trainable params: 10,510,337\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "1/1 [==============================] - 7s 7s/step - loss: 0.6942 - binary_accuracy: 0.5000 - val_loss: 8.6647 - val_binary_accuracy: 0.5000\n",
      "{'loss': [0.6942470073699951], 'binary_accuracy': [0.5], 'val_loss': [8.664695739746094], 'val_binary_accuracy': [0.5]}\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 38ms/step - loss: 8.6647 - binary_accuracy: 0.5000\n",
      "Metrics: 8.664695739746094 0.5\n"
     ]
    }
   ],
   "source": [
    "def train_find_matches(\n",
    "    x_train: np.ndarray,\n",
    "    y_train: np.ndarray,\n",
    "    x_test: np.ndarray,\n",
    "    y_test: np.ndarray,\n",
    "    x_val: np.ndarray,\n",
    "    y_val: np.ndarray,\n",
    "    learning_rate: float = 0.0001,\n",
    "    batch_size: int = 30,\n",
    "    epochs: int = 1,\n",
    "    dropout: float = 0.01,\n",
    ") -> typing.Tuple[keras.models.Model, float, float]:\n",
    "    \"\"\"\n",
    "    Trains a classifier to find matches.\n",
    "    \"\"\"\n",
    "    print(\"Input:\", x_train.shape)\n",
    "    \n",
    "    input_layer: keras.layers.Input = keras.layers.Input(shape=(x_train.shape[1], x_train.shape[2]))\n",
    "    flatten: keras.layers.Flatten = keras.layers.Flatten()(input_layer)\n",
    "\n",
    "    dense1: keras.layers.Dense = keras.layers.Dense(4096, activation='relu')(flatten)\n",
    "    dropout1: keras.layers.Dropout = keras.layers.Dropout(dropout)(dense1)\n",
    "\n",
    "    dense2: keras.layers.Dense = keras.layers.Dense(2048, activation='relu')(dropout1)\n",
    "    dropout2: keras.layers.Dropout = keras.layers.Dropout(dropout)(dense2)\n",
    "\n",
    "    dense3: keras.layers.Dense = keras.layers.Dense(1024, activation='relu')(dropout2)\n",
    "    dropout3: keras.layers.Dropout = keras.layers.Dropout(dropout)(dense3)\n",
    "\n",
    "    dense4: keras.layers.Dense = keras.layers.Dense(514, activation='relu')(dropout3)\n",
    "    dropout4: keras.layers.Dropout = keras.layers.Dropout(dropout)(dense3)\n",
    "\n",
    "    out: keras.layers.Dense = keras.layers.Dense(1, activation='sigmoid')(dropout4)\n",
    "\n",
    "    model: keras.Model = keras.Model(inputs=[input_layer, ], outputs=out)\n",
    "\n",
    "    model.compile(\n",
    "        optimizer=keras.optimizers.RMSprop(learning_rate=learning_rate),\n",
    "        loss=keras.losses.BinaryCrossentropy(),\n",
    "        metrics=[\n",
    "            keras.metrics.BinaryAccuracy()\n",
    "        ],\n",
    "    )\n",
    "\n",
    "    print(model.summary())\n",
    "    \n",
    "    history: keras.callbacks.History = model.fit(\n",
    "        x_train,\n",
    "        y_train,\n",
    "        batch_size=batch_size,\n",
    "        epochs=epochs,\n",
    "        validation_data=(x_test, y_test),\n",
    "        callbacks=[\n",
    "            keras.callbacks.EarlyStopping(monitor='val_loss', patience=3),\n",
    "        ]\n",
    "    )\n",
    "\n",
    "    print(history.history)\n",
    "\n",
    "    # Summarize history of accuracy.\n",
    "    plt.plot(history.history['binary_accuracy'])\n",
    "    plt.plot(history.history['val_binary_accuracy'])\n",
    "    plt.title('Model accuracy')\n",
    "    plt.ylabel('Accuracy')\n",
    "    plt.xlabel('Epoch')\n",
    "    plt.legend(['train', 'test'], loc='upper left')\n",
    "    plt.show()\n",
    "    \n",
    "    # Summarize history for loss.\n",
    "    plt.plot(history.history['loss'])\n",
    "    plt.plot(history.history['val_loss'])\n",
    "    plt.title('Model loss')\n",
    "    plt.ylabel('Loss')\n",
    "    plt.xlabel('Epoch')\n",
    "    plt.legend(['train', 'test'], loc='upper left')\n",
    "    plt.show()\n",
    "\n",
    "    loss, accuracy = model.evaluate(x_val, y_val, batch_size=128)\n",
    "    print(\"Metrics:\", loss, accuracy)\n",
    "    \n",
    "    return model, loss, accuracy\n",
    "\n",
    "\n",
    "x: np.array = np.array([[[0, 1], [0, 1]], [[0, 1], [0, 1]]])\n",
    "y: np.array = np.array([0, 1])\n",
    "sample_model, sample_loss, sample_accuracy = train_find_matches(\n",
    "    x_train=x,\n",
    "    y_train=y,\n",
    "    x_test=x,\n",
    "    y_test=y,\n",
    "    x_val=x,\n",
    "    y_val=y,\n",
    "    epochs=1,\n",
    "    learning_rate=0.01,\n",
    "    batch_size=50,\n",
    ")\n",
    "assert isinstance(sample_model, keras.models.Model)\n",
    "assert isinstance(sample_loss, float)\n",
    "assert isinstance(sample_accuracy, float)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input: (1, 2, 2)\n",
      "Predictions: [[0.97551405]]\n",
      "Input: (2, 2, 2)\n",
      "Predictions: [[0.97551405]\n",
      " [0.97551405]]\n",
      "Input: (3, 2, 2)\n",
      "Predictions: [[0.9755141]\n",
      " [0.9755141]\n",
      " [0.9755141]]\n",
      "Input: (1, 2, 2)\n",
      "Predictions: [[0.97551405]]\n",
      "Confidence: [[ True]]\n"
     ]
    }
   ],
   "source": [
    " def predict_find_matches(\n",
    "    model: keras.models.Model,\n",
    "    records: typing.Union[np.ndarray, list],\n",
    "    confidence: typing.Optional = None,\n",
    " ) -> np.ndarray:\n",
    "    \"\"\"\n",
    "    Uses a trained model to find matches.\n",
    "    \"\"\"\n",
    "    if isinstance(records, list):\n",
    "        records: np.ndarray = vec2matrix(df2vec(pd.DataFrame(data=records, columns=records[0].keys())))\n",
    "    print(\"Input:\", records.shape)\n",
    "    predictions: list = model.predict(records)\n",
    "    print(\"Predictions:\", predictions)\n",
    "    if confidence is not None:\n",
    "        predictions = predictions >= confidence\n",
    "        print(\"Confidence:\", predictions)\n",
    "    return predictions\n",
    "\n",
    "\n",
    "assert predict_find_matches(model=sample_model, records=np.zeros((1, 2, 2))).shape == (1, 1)\n",
    "assert predict_find_matches(model=sample_model, records=np.zeros((2, 2, 2))).shape == (2, 1)\n",
    "assert predict_find_matches(model=sample_model, records=np.zeros((3, 2, 2))).shape == (3, 1)\n",
    "assert predict_find_matches(model=sample_model, records=np.zeros((1, 2, 2)), confidence=0)[0][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib64/python3.9/site-packages/sklearn/metrics/_classification.py:1248: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/usr/local/lib64/python3.9/site-packages/sklearn/metrics/_classification.py:1248: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/usr/local/lib64/python3.9/site-packages/sklearn/metrics/_classification.py:1248: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "       False       0.50      1.00      0.67         1\n",
      "        True       0.00      0.00      0.00         1\n",
      "\n",
      "    accuracy                           0.50         2\n",
      "   macro avg       0.25      0.50      0.33         2\n",
      "weighted avg       0.25      0.50      0.33         2\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAATEAAAEzCAYAAABZrTRjAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8rg+JYAAAACXBIWXMAAAsTAAALEwEAmpwYAAAS3klEQVR4nO3df5BdZX3H8ffXKDU/SHAIBPKDNJFgS2kZfgWBAYJCCDhj7NixoE6rottRI9CKhk4tpMHpVEvpyBisKUXqLyLVqcm0aJCfoQZowtARE7Nk2UwgIUyKIcEktCHut3/she5ZNnt3b/bm7rN5v5gzk3vOuU+eMbMfv89znudsZCaSVKo3tboDknQwDDFJRTPEJBXNEJNUNENMUtEMMUlFM8QkHRIRcUdEbI+Inx/gekTErRHRERE/i4jTB9KuISbpULkTmNfP9cuAWbWjDfjaQBo1xCQdEpm5CtjRzy3zgW9mt8eAoyLi+HrtGmKShospwHM9Pm+pnevXm5vWnZpXX+x0X1PBRk8+v9VdUIP279sajXyv0Z/ZI455+5/QPQx8zdLMXNpIW4PR9BCTVJiuXzf0tVpgHUxobQWm9fg8tXauXw4nJVVlV2PHwVsB/FHtKeU7gV2Zua3el6zEJFV1DUkgvUFE3AXMASZGxBbgRuAtAJn5D8A9wOVAB7AX+OhA2jXEJFXk0FRVfbSbV9a5nsCnB9uuISapqkmVWLMYYpKqmlSJNYshJqmqwaeTrWKISaoqrBJziYWkolmJSapyYl9SyZq1xKJZDDFJVVZikopmJSapaC6xkFQ0KzFJRXNOTFLRrMQkFc1KTFLJMp3Yl1Qyh5OSiuZwUlLRrMQkFc3FrpKKZiUmqWiFzYn5UkRJRbMSk1TlcFJS0QobThpikqoMMUklc9uRpLJZiUkqmhP7kopmJSapaFZikopmJSapaFZikopmJSapaIaYpKI5nJRUNCsxSUWzEpNUtMIqMV+KKKloVmKSqhxOSipaYcNJQ0xSlSEmqWiZre7BoBhikqqsxCQVzRCTVLTCnk66TkxSVVdXY8cARMS8iGiPiI6IuL6P6ydExIMR8WRE/CwiLq/XpiEmqSqzsaOOiBgFLAEuA04GroyIk3vd9gXg7sw8DbgCuK1euw4nJVU1b05sNtCRmZ0AEbEMmA+s73FPAuNrf54APF+vUUNMUlWDIRYRbUBbj1NLM3Npj89TgOd6fN4CnN2rmUXAvRHxGWAscHG9v9cQk1TV4MR+LbCW1r2xf1cCd2bm30XEOcC3IuKUzAN3yhCTVJFdTVvsuhWY1uPz1Nq5nq4C5gFk5qMR8VZgIrD9QI06sS+pqnlPJ9cAsyJiRkQcQffE/Ype9zwLvBsgIn4beCvw3/01aiUmqapJ68Qyc39ELABWAqOAOzJzXUQsBtZm5grgs8A/RsSf0j3J/5HM/h99GmKSqpo3nCQz7wHu6XXuhh5/Xg+cN5g2DbFBuvfBR1j75FNs2NhJe0cne/a+wnvmXsSXbvx8q7umAZoy5XgW3Xgdl86dw9FHv41t27azfMVKbvriLezcuavV3dMgGWKD9PU7l9He0cmY0aOZdOxENm1+rv6XNGzMnDmdRx5ezqRJx7B8xY9pb+/grDNP45qrP86ll87hggvfx44dL7W6m63l3smRbeHVbUw6diInTJ3Mmief4mOfWdjqLmkQvnrrXzNp0jFcc+0XWHLbN14/f/OXb+Taa9u4afFCPr3gDbthDi+FhZhPJwdp9hmnMn3aFCKi1V3RIM2cOZ25c+ewadOz3Pa1OyvXFi2+md279/DhD72fMWNGt6aDw0WTth01iyGmw8acC88F4Cf3raL3A6/du/ewevUaxo4dwzvPPqMV3Rs+mrgBvBnqhlhE/FZELIyIW2vHwtr6Dako7zjp7QBs3NjZ5/WNHZsAmDVr5iHr07DUlY0dLdJviEXEQmAZEMB/1o4A7urrNRrScDZ+wpEA7Nr1cp/XX375VwAcddT4Pq8fNrKrsaNF6k3sXwX8Tma+2vNkRNwCrAP+plkdk9QiLayqGlFvONkFTO7j/PG1a32KiLaIWBsRa2//5l0H0z9pyLy8q7vSmjCh70pr/PjuSm3nzr4rtcNFdnU1dLRKvUrsWuD+iNjI/79C4wTgRGDBgb7Uczf7qy92lhXrGrHan34GOPCc16wTZwAHnjM7bBRWifUbYpn544g4ie6XmU2pnd4KrMnMXze7c9JQeujh1QBccvEFRETlCeW4cWM599yz2LNnL489/kSrujg8jLR37GdmV2Y+lpk/qB2PGWAqUWfnZu699yFmzDiBT33yI5Vri264jnHjxvLt7/yAvXtfaU0Hh4vCnk5GnQ3iB22kDSfvX7WaB1Y9CsCLO17ip48/wdTJx3HGqacA3U+2PrfgE63s4pAaPfn8VndhSPXedrRhw0Zmn3U6F110Hu1PP8P5F8wfMduO9u/b2tCK7D2LrmzoZ3bsortasgLcbUeDtGFjJ8t/dF/l3JbnX2DL8y8AMPm4Y0dUiI00nZ2bOfucy1/fAH7ZvHexbdt2vnLr7W4Af01hc2JWYurXSKvEDicNV2J/+YHGKrGb7rYSkzQMFFaJGWKSKlq55qsRbgCXVDQrMUlVDiclFc0Qk1S0wlbsG2KSqqzEJJWsib8BvCkMMUlVhpikohW2TswQk1RlJSapaIaYpJI1+6UQQ80Qk1RlJSapaIaYpJK5TkxS2QwxSUUra5mYISapyuGkpLIVFmK+2VVS0azEJFU5JyapZM6JSSqblZikklmJSSqblZikkhX2e0IMMUm9GGKSSmYlJqlshYWYK/YlVWRXY8dARMS8iGiPiI6IuP4A93wgItZHxLqI+G69Nq3EJFU0azgZEaOAJcAlwBZgTUSsyMz1Pe6ZBfw5cF5mvhQRx9Zr1xCTVNHEObHZQEdmdgJExDJgPrC+xz2fAJZk5ksAmbm9XqMOJyVVZTR21DcFeK7H5y21cz2dBJwUET+NiMciYl69Rq3EJFU0WolFRBvQ1uPU0sxcOshm3gzMAuYAU4FVEfG7mbmzvy9I0uuya0BV1Ru/1x1Y/YXWVmBaj89Ta+d62gI8npmvApsi4mm6Q23NgRp1OCmpoolPJ9cAsyJiRkQcAVwBrOh1zw/prsKIiIl0Dy87+2vUEJN0SGTmfmABsBL4BXB3Zq6LiMUR8d7abSuBX0bEeuBB4HOZ+cv+2o1m/7bfV1/sLGtLvCpGTz6/1V1Qg/bv29rQuHDrOe9q6Gd2yqMPNDYOPUjOiUmqcNuRpKI1OrHfKoaYpIomzzANOUNMUoWVmKSiGWKSiuZwUlLRrMQkFS0Htpl72DDEJFW4TkxS0bqsxCSVzOGkpKI5sS+paC6xkFQ0KzFJRSttYt+XIkoqmpWYpAqfTkoqmhP7kopW2pyYISapwuGkpKI5nJRUNIeTkormcFJS0azEJBWtsCkxQ0xSlZWYpKI5JyapaIW9ndoQk1SVWIlJKlhXYTP7hpikii4rMUklK2046UsRJRXNSkxShU8nJRWttOGkISapwkpMUtEMMUlFczgpqWiF/e5cQ0xSlYtdJRWtsF1HhpikKif2JRWtKxxOSiqYw0lJRSttOOkGcEkVXdHYMRARMS8i2iOiIyKu7+e+90dERsSZ9dq0EpNU0awlFhExClgCXAJsAdZExIrMXN/rviOBa4DHB9KulZikimzwGIDZQEdmdmbmPmAZML+P+24CvgT8z0AaNcQkVTRxODkFeK7H5y21c6+LiNOBaZn57wPtryEmaUhERFtErO1xtA3y+28CbgE+O5jvOScmqaLRp5OZuRRY2s8tW4FpPT5PrZ17zZHAKcBD0b1W7ThgRUS8NzPXHqhRQ0xSRRPXia0BZkXEDLrD6wrgg6//vZm7gImvfY6Ih4Dr+gswMMQk9dKst1hk5v6IWACsBEYBd2TmuohYDKzNzBWNtGuISapo5mLXzLwHuKfXuRsOcO+cgbRpiEmqKG3FviEmqSLL2v9tiEmqshKTVDRDTFLRfBWPpKL5i0IkFc3hpKSiGWKSiuacmKSiOScmqWgOJyUVzeGkpKJ1FRZjvtlVUtGsxCRVOCcmqWhlDSYNMUm9WIlJKprrxCQVrbSnk4aYpIqyIswQk9SLc2KSiuZwUlLRyoowQ0xSLw4nJRXN4aSkopUVYYaYpF4cTkoqWhZWixlig3Tvg4+w9smn2LCxk/aOTvbsfYX3zL2IL934+VZ3TQM0ZcrxLLrxOi6dO4ejj34b27ZtZ/mKldz0xVvYuXNXq7vXclZiI9zX71xGe0cnY0aPZtKxE9m0+blWd0mDMHPmdB55eDmTJh3D8hU/pr29g7POPI1rrv44l146hwsufB87drzU6m62lBP7I9zCq9uYdOxETpg6mTVPPsXHPrOw1V3SIHz11r9m0qRjuObaL7Dktm+8fv7mL9/Itde2cdPihXx6wfUt7KEGyze7DtLsM05l+rQpRBS21V/MnDmduXPnsGnTs9z2tTsr1xYtvpndu/fw4Q+9nzFjRremg8NENni0iiGmw8acC88F4Cf3rSKz+mO3e/ceVq9ew9ixY3jn2We0onvDRhfZ0NEqhpgOG+846e0AbNzY2ef1jR2bAJg1a+Yh69Nw1NXg0SoNh1hEfHQoOyI12/gJRwKwa9fLfV5/+eVfAXDUUeMPWZ+Go2zwv1Y5mErsr4asF5KGjRFViUXEzw5wPAVM6ud7bRGxNiLW3v7Nu4a801IjXt7VXWlNmNB3pTV+fHeltnNn35Xa4aK0SqzeEotJwKVA74UzAaw+0JcycymwFODVFzvLWnSiEav96WeAA895zTpxBnDgObPDxUhb7PpvwLjM/K/eFyLioWZ0SGqWhx7u/v/dSy6+gIioPKEcN24s5557Fnv27OWxx59oVReHha4sq+7odziZmVdl5n8c4NoHm9MlqTk6Ozdz770PMWPGCXzqkx+pXFt0w3WMGzeWb3/nB+zd+0prOjhMlLZOLHqvlxlqI204ef+q1Tyw6lEAXtzxEj99/AmmTj6OM049Beh+svW5BZ9oZReH1OjJ57e6C0Oq97ajDRs2Mvus07noovNof/oZzr9g/ojZdrR/39aGVmR/cPrvN/Qz+93N/9qSFeBuOxqkDRs7Wf6j+yrntjz/AluefwGAyccdO6JCbKTp7NzM2edc/voG8MvmvYtt27bzlVtvdwN4TWlvsbASU79GWiV2OGm0EvvD6e9r6Gf2e5t/aCUmqfV8i4WkopU2nDTEJFWUtk7MDeCSKjKzoWMgImJeRLRHREdEvOHFbRHxZxGxvrYz6P6ImF6vTUNM0iEREaOAJcBlwMnAlRFxcq/bngTOzMzfA74PfLleu4aYpIomvk9sNtCRmZ2ZuQ9YBszveUNmPpiZe2sfHwOm1mvUEJNU0ehbLHq++KF2tPVqegrQ85dSbKmdO5CrgB/V668T+5IqGn062fPFDwcrIj4MnAlcWO9eQ0xSRRPXiW0FpvX4PLV2riIiLgb+ArgwM/+3XqOGmKSKJu7iWQPMiogZdIfXFUDlRRIRcRrwdWBeZm4fSKOGmKSKZq0Ty8z9EbEAWAmMAu7IzHURsRhYm5krgL8FxgH/UvuNYs9m5nv7a9cQk1TRzBX7mXkPcE+vczf0+PPFg23TEJNU4d5JSUVr9ptthpohJqnCSkxS0XyLhaSilfaLQgwxSRVlRZghJqkX58QkFc0Qk1S00pZY+CoeSUWzEpNU4XBSUtFcJyapaKXNiRlikiocTkoqmpWYpKJZiUkqmhP7kormBnBJRbMSk1Q0KzFJRbMSk1Q0KzFJRbMSk1Q0KzFJRbMSk1S0zK5Wd2FQfCmipKJZiUmqcO+kpKL5FgtJRbMSk1Q0KzFJRXOdmKSiuU5MUtEcTkoqmhP7kopmJSapaE7sSyqalZikojknJqloVmKSiuacmKSiudhVUtGsxCQVrbQ5Md/sKqlohpikimzwv4GIiHkR0R4RHRFxfR/XfyMivle7/nhE/Ga9Ng0xSRWZ2dBRT0SMApYAlwEnA1dGxMm9brsKeCkzTwT+HvhSvXYNMUkVzQoxYDbQkZmdmbkPWAbM73XPfOCfa3/+PvDuiIj+GjXEJFVkg8cATAGe6/F5S+1cn/dk5n5gF3B0f402/enkWybO7DdFSxcRbZm5tNX9aJb9+7a2ugtNNdL//Rqxf9/Whn5mI6INaOtxaumh+N/WSuzgtdW/RcOY/35DJDOXZuaZPY7eAbYVmNbj89TauT7viYg3AxOAX/b39xpikg6VNcCsiJgREUcAVwAret2zAvjj2p//AHgg60y4udhV0iGRmfsjYgGwEhgF3JGZ6yJiMbA2M1cA/wR8KyI6gB10B12/orTVucONcypl89+vfIaYpKI5JyapaIbYQai3hULDV0TcERHbI+Lnre6LDo4h1qABbqHQ8HUnMK/VndDBM8QaN5AtFBqmMnMV3U+/VDhDrHED2UIhqckMMUlFM8QaN5AtFJKazBBr3EC2UEhqMkOsQbXXhLy2heIXwN2Zua61vdJARcRdwKPAOyJiS0Rc1eo+qTGu2JdUNCsxSUUzxCQVzRCTVDRDTFLRDDFJRTPEJBXNEJNUNENMUtH+D3ZD/OFpO560AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 360x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def error_analysis(\n",
    "    model: keras.models.Model,\n",
    "    x_test: np.array,\n",
    "    y_test: np.array,\n",
    "    confidence: float\n",
    ") -> None:\n",
    "    \"\"\"\n",
    "    Performs error analysis on a prediction model.\n",
    "    \"\"\"\n",
    "    y_pred: np.ndarray = np.array(model.predict(x_test))\n",
    "\n",
    "    # ROC Curve\n",
    "    fpr, tpr, thresholds = roc_curve(y_test, y_pred)\n",
    "    plt.plot(fpr, tpr) \n",
    "    plt.axis([0, 1, 0, 1]) \n",
    "    plt.xlabel('False Positive Rate') \n",
    "    plt.ylabel('True Positive Rate') \n",
    "    plt.show()\n",
    "        \n",
    "    # Confusion Matrix\n",
    "    matrix = confusion_matrix(y_test > confidence, y_pred > confidence)\n",
    "    print(classification_report(y_test > confidence, y_pred > confidence))\n",
    "    cm: np.array = confusion_matrix(y_test > confidence, y_pred > confidence)\n",
    "    cm: pd.DataFrame = pd.DataFrame(cm, range(2), range(2))\n",
    "    plt.figure(figsize = (5, 5))\n",
    "    sns.heatmap(cm, annot=True, annot_kws={\"size\": 20})\n",
    "    plt.show()\n",
    "    \n",
    "assert error_analysis(\n",
    "    model=sample_model,\n",
    "    confidence=0.3,\n",
    "    x_test=np.array([[[0, 1], [0, 1]], [[0, 1], [0, 1]]]),\n",
    "    y_test=np.array([0, 1]),\n",
    ") is None"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Companies Find Matches"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Companies: (10000, 17)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "raw_companies_dataframe: pd.DataFrame = generate_random_companies(\n",
    "    total=9900,\n",
    "    nulls=.01,\n",
    "    mutability=.01,\n",
    "    matches=.5,\n",
    "    relevance=50,\n",
    "    feedback=[\n",
    "        (\n",
    "            {\n",
    "                \"city\": \"San Francisco\",\n",
    "                \"country\": \"USA\",\n",
    "                \"company_name\": \"Airbnb\",\n",
    "                \"phone\": \"120-403-4032\",\n",
    "                \"street_address\": \"960 Air Ave.\",\n",
    "                \"website\": \"www.airbnb.com\",\n",
    "                \"linkedin_url\": \"https://www.linkedin.com/u/airbnb\",\n",
    "                \"facebook_url\": \"https://www.facebook.com/user/airbnb\",\n",
    "            },\n",
    "            {\n",
    "                \"city\": \"San Francisco\",\n",
    "                \"country\": \"USA\",\n",
    "                \"company_name\": \"airbnb\",\n",
    "                \"phone\": \"120-403-4032\",\n",
    "                \"street_address\": \"960 Air Ave.\",\n",
    "                \"website\": \"www.airbnb.com.ar\",\n",
    "                \"linkedin_url\": \"https://www.linkedin.com/u/airbnb\",\n",
    "                \"facebook_url\": \"https://www.facebook.com/user/airbnb\",\n",
    "            },\n",
    "            1\n",
    "        ),\n",
    "        (\n",
    "            {\n",
    "                \"city\": \"San Francisco\",\n",
    "                \"country\": \"USA\",\n",
    "                \"company_name\": \"Airbnb\",\n",
    "                \"phone\": \"120-403-4032\",\n",
    "                \"street_address\": \"960 Air Ave.\",\n",
    "                \"website\": \"www.airbnb.com\",\n",
    "                \"linkedin_url\": \"https://www.linkedin.com/u/airbnb\",\n",
    "                \"facebook_url\": \"https://www.facebook.com/user/airbnb\",\n",
    "            },\n",
    "            {\n",
    "                \"city\": \"Mexico\",\n",
    "                \"country\": \"Mexico\",\n",
    "                \"company_name\": \"Taco Bell\",\n",
    "                \"phone\": \"543-543-1231\",\n",
    "                \"street_address\": \"50 Something St.\",\n",
    "                \"website\": \"www.taco-bell.com\",\n",
    "                \"linkedin_url\": \"https://www.linkedin.com/u/taco\",\n",
    "                \"facebook_url\": \"https://www.facebook.com/user/taco\",  \n",
    "            },\n",
    "            0\n",
    "        ),  \n",
    "    ]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>city#1</th>\n",
       "      <th>country#1</th>\n",
       "      <th>company_name#1</th>\n",
       "      <th>phone#1</th>\n",
       "      <th>street_address#1</th>\n",
       "      <th>website#1</th>\n",
       "      <th>linkedin_url#1</th>\n",
       "      <th>facebook_url#1</th>\n",
       "      <th>city#2</th>\n",
       "      <th>country#2</th>\n",
       "      <th>company_name#2</th>\n",
       "      <th>phone#2</th>\n",
       "      <th>street_address#2</th>\n",
       "      <th>website#2</th>\n",
       "      <th>linkedin_url#2</th>\n",
       "      <th>facebook_url#2</th>\n",
       "      <th>match</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Gmea Loaiom</td>\n",
       "      <td>Zierl Oaaw</td>\n",
       "      <td>Asugreileg Uncuooprgso</td>\n",
       "      <td>454-387-2793</td>\n",
       "      <td>788 Aauocm Amoio, Cuuoec St.</td>\n",
       "      <td>www.retgioig.net</td>\n",
       "      <td>www.linkedin.com/u/keporploo</td>\n",
       "      <td>https:///www.facebook.com/profile/soepclteeo</td>\n",
       "      <td>Gmea Loaiom</td>\n",
       "      <td>Zierl Oaaw</td>\n",
       "      <td>Asugreileg Uncuooprgso</td>\n",
       "      <td>454-387-2793</td>\n",
       "      <td>788 Aauocm Amoio, CuuoeclSt.</td>\n",
       "      <td>www.retgioig.net</td>\n",
       "      <td>www.linkedinacom/u/keporploo</td>\n",
       "      <td>https:///www.facebook.com/profile/soepclteeo</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Pueomg Ric</td>\n",
       "      <td>Rputesoicoai</td>\n",
       "      <td>Riuaucitoi Ucrozrigelr</td>\n",
       "      <td>765-891-3829</td>\n",
       "      <td>733 Unoor Rree, Irig St.</td>\n",
       "      <td>www.simaezuemogpu.com</td>\n",
       "      <td>www.linkedin.com/u/isjoeaii</td>\n",
       "      <td>https:///www.facebook.com/profile/gaicgpiarr</td>\n",
       "      <td>Pueomg Ric</td>\n",
       "      <td>Rputesoicoai</td>\n",
       "      <td>Riuaucitoi Ucrozrigelrc</td>\n",
       "      <td>765-891-3829</td>\n",
       "      <td>733 Unoor Rree, Irig St.</td>\n",
       "      <td>www.simaezuemogpu.com</td>\n",
       "      <td>www.linaedin.com/u/isjoeaii</td>\n",
       "      <td>https:///www.facebook.com/profile/gaicgpiarr</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Aeeupiiee</td>\n",
       "      <td>Uixo Eeac</td>\n",
       "      <td>Lljegrop Asoaurug</td>\n",
       "      <td>654-146-5629</td>\n",
       "      <td>974 Grn Ouaeum, Sepru St.</td>\n",
       "      <td>www.cctissritoagegg.net</td>\n",
       "      <td>www.linkedin.com/u/rerpmalrr</td>\n",
       "      <td>https:///www.facebook.com/profile/seautguesvge</td>\n",
       "      <td>Iiuzsauloga</td>\n",
       "      <td>Ntrguu Aoromi</td>\n",
       "      <td>Moonrocn Rrlegais</td>\n",
       "      <td>466-938-6623</td>\n",
       "      <td>46 Hmp Mnsf, Opvoep St.</td>\n",
       "      <td>www.usnoiue.org</td>\n",
       "      <td>www.linkedin.com/u/gtuccreiuu</td>\n",
       "      <td>https:///www.facebook.com/profile/ezmcrursuueee</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Ymc Oanspa</td>\n",
       "      <td>Euaaim Cieo</td>\n",
       "      <td>Teoieglt Inaitsnerol</td>\n",
       "      <td>181-575-7745</td>\n",
       "      <td>791 Eoaig Oeenl, Pmpeup Av.</td>\n",
       "      <td>www.aoceceslp.org</td>\n",
       "      <td>www.linkedin.com/u/ioorbcoeait</td>\n",
       "      <td>https:///www.facebook.com/profile/lilbecjgtnceeiu</td>\n",
       "      <td>Ymc Oanspa</td>\n",
       "      <td>Euaaim Cieo</td>\n",
       "      <td>Teoieglt Inaitsnerol</td>\n",
       "      <td>181-575-7745</td>\n",
       "      <td>791 Eoaig Oeenl, Pmpeup Av.</td>\n",
       "      <td>www.aoceceslp.org</td>\n",
       "      <td>www.linkedin.com/u/ioorbcoeait</td>\n",
       "      <td>https:///www.facebook.com/profile/lilbecjgtnceeiu</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Aso Ogao</td>\n",
       "      <td>Otiaepnas</td>\n",
       "      <td>Eeorpicg Aiiauirsurri</td>\n",
       "      <td>892-996-7926</td>\n",
       "      <td>413 Caranmu Pirtt, Euma Av.</td>\n",
       "      <td>www.irlromdmmriu.org</td>\n",
       "      <td>www.linkedin.com/u/ggpuuteranop</td>\n",
       "      <td>https:///www.facebook.com/profile/gneoerin</td>\n",
       "      <td>Oiotoclm</td>\n",
       "      <td>Igoui Patal</td>\n",
       "      <td>Maepusesen Nuuielpp</td>\n",
       "      <td>706-721-8588</td>\n",
       "      <td></td>\n",
       "      <td>www.eioimrumttsae.com</td>\n",
       "      <td>www.linkedin.com/u/acnopragei</td>\n",
       "      <td>https:///www.facebook.com/profile/treimara</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Norea Ouloo</td>\n",
       "      <td>Rrzm Lpln</td>\n",
       "      <td>Moetoseusrpf Gktjcuoe</td>\n",
       "      <td>106-840-9451</td>\n",
       "      <td>214 Ieirta Neuuep, Nceo Av.</td>\n",
       "      <td>www.agnaarimoaozp.net</td>\n",
       "      <td>www.linkedin.com/u/ceeuimorepsro</td>\n",
       "      <td>https:///www.facebook.com/profile/qtausuaciegit</td>\n",
       "      <td>Norea Ouloo</td>\n",
       "      <td>Rrzm Lpln</td>\n",
       "      <td>Moetoseusrpf Gktjcuoe</td>\n",
       "      <td>106-840-9451</td>\n",
       "      <td>214 Ieirta Neuuep, Nceo Av.</td>\n",
       "      <td>www.agnaarimoaozp.net</td>\n",
       "      <td>www.linkedin.com/u/ceeuimorepsro</td>\n",
       "      <td>https:///www.facebook.com/profile/qtausuaciegit</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Slaeeluu</td>\n",
       "      <td>Aausrnpo</td>\n",
       "      <td>Arcopuio Roonmtasce</td>\n",
       "      <td>543-946-8389</td>\n",
       "      <td>655 Uea Pertnr, Uage St.</td>\n",
       "      <td>www.iaeaerrmuulvuin.org</td>\n",
       "      <td>www.linkedin.com/u/eciggoietm</td>\n",
       "      <td>https:///www.facebook.com/profile/goeoeesoonerae</td>\n",
       "      <td>Slaeeluu</td>\n",
       "      <td>Aausunpo</td>\n",
       "      <td>Arcopuio Rooomtasce</td>\n",
       "      <td>543-946-8389</td>\n",
       "      <td>655 Uea Pertnr, Uage St.</td>\n",
       "      <td>www.iaeaerrmuulvuin.org</td>\n",
       "      <td>www.linkedin.com/u/ecrggoietm</td>\n",
       "      <td>https:///www.facebook.com/profile/goeoeesoonerae</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Pnulesolo</td>\n",
       "      <td>Aumoonec</td>\n",
       "      <td>Hccfrieulnms Oupeiuvmeurn</td>\n",
       "      <td>437-897-1683</td>\n",
       "      <td>755 Aaoiaar Meee, Uaoi Av.</td>\n",
       "      <td>www.uermlpsiolgpu.net</td>\n",
       "      <td>www.linkedin.com/u/ggiyppcsanmroum</td>\n",
       "      <td>https:///www.facebook.com/profile/cgnrouirseaenen</td>\n",
       "      <td>Pnulesolo</td>\n",
       "      <td>Aumoonec</td>\n",
       "      <td>Hccfrieulnms Oupeiuvmeurn</td>\n",
       "      <td>437-897-1683</td>\n",
       "      <td>755 Aaoiaar Meee, Uaoi Av.</td>\n",
       "      <td>www.uermlssiolgpu.net</td>\n",
       "      <td>www.linkedin.com/u/ggiyppcsanmroum</td>\n",
       "      <td>https:///www.facebook.com/profile/cgnrouirseaenen</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Emt Uap</td>\n",
       "      <td>Mlra Spe</td>\n",
       "      <td>Rilnotlrn Nlduacau</td>\n",
       "      <td>773-644-6572</td>\n",
       "      <td>787 Ucf Ntid, Orsri St.</td>\n",
       "      <td>www.isulopparmtmn.org</td>\n",
       "      <td>www.linkedin.com/u/eiuaiogolprtgup</td>\n",
       "      <td>https:///www.facebook.com/profile/rumiingotieee</td>\n",
       "      <td>Giiuorno</td>\n",
       "      <td>Ieugausooo</td>\n",
       "      <td>Iluocela Onurugsugu</td>\n",
       "      <td></td>\n",
       "      <td>202 Rceiraec Ruta, Llcou Av.</td>\n",
       "      <td>www.auiignrnthr.com</td>\n",
       "      <td>www.linkedin.com/u/oreiragauotgme</td>\n",
       "      <td>https:///www.facebook.com/profile/naocccuoulns</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Plgsmoao</td>\n",
       "      <td>Eaocriniocgu</td>\n",
       "      <td>Tramturnape Ueunygiirmro</td>\n",
       "      <td>700-269-4169</td>\n",
       "      <td>495 Til Tocur, Eoiuc St.</td>\n",
       "      <td>www.aoratlibmiera.net</td>\n",
       "      <td>www.linkedin.com/u/oaanunuaslaaaau</td>\n",
       "      <td>https:///www.facebook.com/profile/aptodonmo</td>\n",
       "      <td>Uorm Smors</td>\n",
       "      <td>Gcpo Ertrcr</td>\n",
       "      <td>Oiaqsrfg Oruplarr</td>\n",
       "      <td>349-431-1318</td>\n",
       "      <td>639 Opis Rrniut, Sireuo Av.</td>\n",
       "      <td>www.ueuciorourve.com</td>\n",
       "      <td>www.linkedin.com/u/lgeuearun</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        city#1     country#1             company_name#1       phone#1  \\\n",
       "0  Gmea Loaiom    Zierl Oaaw     Asugreileg Uncuooprgso  454-387-2793   \n",
       "1   Pueomg Ric  Rputesoicoai     Riuaucitoi Ucrozrigelr  765-891-3829   \n",
       "2    Aeeupiiee     Uixo Eeac          Lljegrop Asoaurug  654-146-5629   \n",
       "3   Ymc Oanspa   Euaaim Cieo       Teoieglt Inaitsnerol  181-575-7745   \n",
       "4     Aso Ogao     Otiaepnas      Eeorpicg Aiiauirsurri  892-996-7926   \n",
       "5  Norea Ouloo     Rrzm Lpln      Moetoseusrpf Gktjcuoe  106-840-9451   \n",
       "6     Slaeeluu      Aausrnpo        Arcopuio Roonmtasce  543-946-8389   \n",
       "7    Pnulesolo      Aumoonec  Hccfrieulnms Oupeiuvmeurn  437-897-1683   \n",
       "8      Emt Uap      Mlra Spe         Rilnotlrn Nlduacau  773-644-6572   \n",
       "9     Plgsmoao  Eaocriniocgu   Tramturnape Ueunygiirmro  700-269-4169   \n",
       "\n",
       "               street_address#1                website#1  \\\n",
       "0  788 Aauocm Amoio, Cuuoec St.         www.retgioig.net   \n",
       "1      733 Unoor Rree, Irig St.    www.simaezuemogpu.com   \n",
       "2     974 Grn Ouaeum, Sepru St.  www.cctissritoagegg.net   \n",
       "3   791 Eoaig Oeenl, Pmpeup Av.        www.aoceceslp.org   \n",
       "4   413 Caranmu Pirtt, Euma Av.     www.irlromdmmriu.org   \n",
       "5   214 Ieirta Neuuep, Nceo Av.    www.agnaarimoaozp.net   \n",
       "6      655 Uea Pertnr, Uage St.  www.iaeaerrmuulvuin.org   \n",
       "7    755 Aaoiaar Meee, Uaoi Av.    www.uermlpsiolgpu.net   \n",
       "8       787 Ucf Ntid, Orsri St.    www.isulopparmtmn.org   \n",
       "9      495 Til Tocur, Eoiuc St.    www.aoratlibmiera.net   \n",
       "\n",
       "                       linkedin_url#1  \\\n",
       "0        www.linkedin.com/u/keporploo   \n",
       "1         www.linkedin.com/u/isjoeaii   \n",
       "2        www.linkedin.com/u/rerpmalrr   \n",
       "3      www.linkedin.com/u/ioorbcoeait   \n",
       "4     www.linkedin.com/u/ggpuuteranop   \n",
       "5    www.linkedin.com/u/ceeuimorepsro   \n",
       "6       www.linkedin.com/u/eciggoietm   \n",
       "7  www.linkedin.com/u/ggiyppcsanmroum   \n",
       "8  www.linkedin.com/u/eiuaiogolprtgup   \n",
       "9  www.linkedin.com/u/oaanunuaslaaaau   \n",
       "\n",
       "                                      facebook_url#1       city#2  \\\n",
       "0       https:///www.facebook.com/profile/soepclteeo  Gmea Loaiom   \n",
       "1       https:///www.facebook.com/profile/gaicgpiarr   Pueomg Ric   \n",
       "2     https:///www.facebook.com/profile/seautguesvge  Iiuzsauloga   \n",
       "3  https:///www.facebook.com/profile/lilbecjgtnceeiu   Ymc Oanspa   \n",
       "4         https:///www.facebook.com/profile/gneoerin     Oiotoclm   \n",
       "5    https:///www.facebook.com/profile/qtausuaciegit  Norea Ouloo   \n",
       "6   https:///www.facebook.com/profile/goeoeesoonerae     Slaeeluu   \n",
       "7  https:///www.facebook.com/profile/cgnrouirseaenen    Pnulesolo   \n",
       "8    https:///www.facebook.com/profile/rumiingotieee     Giiuorno   \n",
       "9        https:///www.facebook.com/profile/aptodonmo   Uorm Smors   \n",
       "\n",
       "       country#2             company_name#2       phone#2  \\\n",
       "0     Zierl Oaaw     Asugreileg Uncuooprgso  454-387-2793   \n",
       "1   Rputesoicoai    Riuaucitoi Ucrozrigelrc  765-891-3829   \n",
       "2  Ntrguu Aoromi          Moonrocn Rrlegais  466-938-6623   \n",
       "3    Euaaim Cieo       Teoieglt Inaitsnerol  181-575-7745   \n",
       "4    Igoui Patal        Maepusesen Nuuielpp  706-721-8588   \n",
       "5      Rrzm Lpln      Moetoseusrpf Gktjcuoe  106-840-9451   \n",
       "6       Aausunpo        Arcopuio Rooomtasce  543-946-8389   \n",
       "7       Aumoonec  Hccfrieulnms Oupeiuvmeurn  437-897-1683   \n",
       "8     Ieugausooo        Iluocela Onurugsugu                 \n",
       "9    Gcpo Ertrcr          Oiaqsrfg Oruplarr  349-431-1318   \n",
       "\n",
       "               street_address#2                website#2  \\\n",
       "0  788 Aauocm Amoio, CuuoeclSt.         www.retgioig.net   \n",
       "1      733 Unoor Rree, Irig St.    www.simaezuemogpu.com   \n",
       "2       46 Hmp Mnsf, Opvoep St.          www.usnoiue.org   \n",
       "3   791 Eoaig Oeenl, Pmpeup Av.        www.aoceceslp.org   \n",
       "4                                  www.eioimrumttsae.com   \n",
       "5   214 Ieirta Neuuep, Nceo Av.    www.agnaarimoaozp.net   \n",
       "6      655 Uea Pertnr, Uage St.  www.iaeaerrmuulvuin.org   \n",
       "7    755 Aaoiaar Meee, Uaoi Av.    www.uermlssiolgpu.net   \n",
       "8  202 Rceiraec Ruta, Llcou Av.      www.auiignrnthr.com   \n",
       "9   639 Opis Rrniut, Sireuo Av.     www.ueuciorourve.com   \n",
       "\n",
       "                       linkedin_url#2  \\\n",
       "0        www.linkedinacom/u/keporploo   \n",
       "1         www.linaedin.com/u/isjoeaii   \n",
       "2       www.linkedin.com/u/gtuccreiuu   \n",
       "3      www.linkedin.com/u/ioorbcoeait   \n",
       "4       www.linkedin.com/u/acnopragei   \n",
       "5    www.linkedin.com/u/ceeuimorepsro   \n",
       "6       www.linkedin.com/u/ecrggoietm   \n",
       "7  www.linkedin.com/u/ggiyppcsanmroum   \n",
       "8   www.linkedin.com/u/oreiragauotgme   \n",
       "9        www.linkedin.com/u/lgeuearun   \n",
       "\n",
       "                                      facebook_url#2  match  \n",
       "0       https:///www.facebook.com/profile/soepclteeo      1  \n",
       "1       https:///www.facebook.com/profile/gaicgpiarr      1  \n",
       "2    https:///www.facebook.com/profile/ezmcrursuueee      0  \n",
       "3  https:///www.facebook.com/profile/lilbecjgtnceeiu      1  \n",
       "4         https:///www.facebook.com/profile/treimara      0  \n",
       "5    https:///www.facebook.com/profile/qtausuaciegit      1  \n",
       "6   https:///www.facebook.com/profile/goeoeesoonerae      1  \n",
       "7  https:///www.facebook.com/profile/cgnrouirseaenen      1  \n",
       "8     https:///www.facebook.com/profile/naocccuoulns      0  \n",
       "9                                                         0  "
      ]
     },
     "execution_count": 128,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "raw_companies_dataframe.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Vectorizing: city#1\n",
      "Vectorizing: country#1\n",
      "Vectorizing: company_name#1\n",
      "Vectorizing: phone#1\n",
      "Vectorizing: street_address#1\n",
      "Vectorizing: website#1\n",
      "Vectorizing: linkedin_url#1\n",
      "Vectorizing: facebook_url#1\n",
      "Vectorizing: city#2\n",
      "Vectorizing: country#2\n",
      "Vectorizing: company_name#2\n",
      "Vectorizing: phone#2\n",
      "Vectorizing: street_address#2\n",
      "Vectorizing: website#2\n",
      "Vectorizing: linkedin_url#2\n",
      "Vectorizing: facebook_url#2\n",
      "Vectorizing: match\n"
     ]
    }
   ],
   "source": [
    "vectorized_companies_dataframe: pd.DataFrame = df2vec(raw_companies_dataframe)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>city#1</th>\n",
       "      <th>country#1</th>\n",
       "      <th>company_name#1</th>\n",
       "      <th>phone#1</th>\n",
       "      <th>street_address#1</th>\n",
       "      <th>website#1</th>\n",
       "      <th>linkedin_url#1</th>\n",
       "      <th>facebook_url#1</th>\n",
       "      <th>city#2</th>\n",
       "      <th>country#2</th>\n",
       "      <th>company_name#2</th>\n",
       "      <th>phone#2</th>\n",
       "      <th>street_address#2</th>\n",
       "      <th>website#2</th>\n",
       "      <th>linkedin_url#2</th>\n",
       "      <th>facebook_url#2</th>\n",
       "      <th>match</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>[-0.17081124, -0.64723957, 0.22311965, 0.11456...</td>\n",
       "      <td>[-0.1733922, -0.29064077, 0.08646101, 0.469208...</td>\n",
       "      <td>[-0.2278746, -0.33084768, -0.6021811, 0.147618...</td>\n",
       "      <td>[0.1663901, 0.4832298, 0.31022692, -0.52979314...</td>\n",
       "      <td>[0.13965015, -0.3673879, 0.4392662, -0.1811918...</td>\n",
       "      <td>[0.8729764, -0.21711689, 0.14693439, 0.1163865...</td>\n",
       "      <td>[0.30312464, -0.3625717, 0.23163374, 0.4615468...</td>\n",
       "      <td>[0.37542623, 0.53012043, 0.40159354, -0.314179...</td>\n",
       "      <td>[-0.17081124, -0.64723957, 0.22311965, 0.11456...</td>\n",
       "      <td>[-0.1733922, -0.29064077, 0.08646101, 0.469208...</td>\n",
       "      <td>[-0.2278746, -0.33084768, -0.6021811, 0.147618...</td>\n",
       "      <td>[0.1663901, 0.4832298, 0.31022692, -0.52979314...</td>\n",
       "      <td>[0.054061692, -0.09831729, 0.62615895, 0.04232...</td>\n",
       "      <td>[0.8729764, -0.21711689, 0.14693439, 0.1163865...</td>\n",
       "      <td>[-0.35564625, 0.0014702752, -0.2596981, -0.141...</td>\n",
       "      <td>[0.37542623, 0.53012043, 0.40159354, -0.314179...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>[0.28466508, -0.6870391, 0.034863576, 0.143425...</td>\n",
       "      <td>[-0.45580024, -0.64020044, -0.94991213, 1.1212...</td>\n",
       "      <td>[-0.077516824, -0.10479821, 0.33265123, -0.063...</td>\n",
       "      <td>[0.10064737, 0.48980132, 0.121493064, -0.52424...</td>\n",
       "      <td>[-0.10337315, -0.25847742, 0.5891691, -0.21212...</td>\n",
       "      <td>[0.01968214, 0.6168068, -0.90221906, 0.0048712...</td>\n",
       "      <td>[-0.456138, -0.53931075, -0.30390802, 0.474316...</td>\n",
       "      <td>[0.21144052, 0.20298834, 0.21815896, -0.486129...</td>\n",
       "      <td>[0.28466508, -0.6870391, 0.034863576, 0.143425...</td>\n",
       "      <td>[-0.45580024, -0.64020044, -0.94991213, 1.1212...</td>\n",
       "      <td>[-0.3959472, -0.06375926, 0.039568715, 0.15786...</td>\n",
       "      <td>[0.10064737, 0.48980132, 0.121493064, -0.52424...</td>\n",
       "      <td>[-0.10337315, -0.25847742, 0.5891691, -0.21212...</td>\n",
       "      <td>[0.01968214, 0.6168068, -0.90221906, 0.0048712...</td>\n",
       "      <td>[-0.6107266, -1.229, 0.017428547, -0.58206856,...</td>\n",
       "      <td>[0.21144052, 0.20298834, 0.21815896, -0.486129...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>[0.008446187, -0.89799815, 0.3153159, -0.01146...</td>\n",
       "      <td>[-0.08009297, -0.70065546, 0.8756641, -0.21270...</td>\n",
       "      <td>[-0.05640249, -0.50657225, 0.198513, 0.1292749...</td>\n",
       "      <td>[0.22372286, 0.7134844, 0.27984577, -0.703583,...</td>\n",
       "      <td>[-0.26451614, -0.3932885, 0.4730599, -0.185341...</td>\n",
       "      <td>[0.19262332, -0.031438783, -0.19407359, 0.1009...</td>\n",
       "      <td>[-0.28189784, -0.4292636, 0.48723325, -0.07348...</td>\n",
       "      <td>[0.5606265, 0.4351116, 0.43130416, -0.25141925...</td>\n",
       "      <td>[-0.21692994, -0.59062845, -0.63720477, 0.0989...</td>\n",
       "      <td>[-0.2784196, -0.5214716, -0.2022338, 0.1992558...</td>\n",
       "      <td>[-0.05473563, -0.11676584, 0.12873732, -0.0655...</td>\n",
       "      <td>[0.12499511, 0.7256371, 0.2397831, -0.56979215...</td>\n",
       "      <td>[-0.23111165, -0.4517533, 0.1346171, -0.219431...</td>\n",
       "      <td>[0.0700393, -0.6345332, 0.08723115, 0.00782650...</td>\n",
       "      <td>[0.3146601, -0.14218765, 0.77169657, -0.107591...</td>\n",
       "      <td>[0.533215, 0.30011246, 0.2952332, -0.20062599,...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>[0.4621147, -0.32601243, 0.14875361, 0.0588712...</td>\n",
       "      <td>[-0.17904884, -0.30857, -0.08816792, 0.2628701...</td>\n",
       "      <td>[0.3672084, -0.39905167, 0.45736057, -0.170849...</td>\n",
       "      <td>[0.15963465, 0.23201807, 0.07888236, -0.746187...</td>\n",
       "      <td>[-0.27435634, -0.34182993, 0.42234477, -0.0363...</td>\n",
       "      <td>[-0.8862241, -0.77771133, -0.3682679, -0.15156...</td>\n",
       "      <td>[-0.24318284, -1.1066128, -0.28148663, -0.1359...</td>\n",
       "      <td>[0.3766895, 0.3671902, 0.2224524, -0.2929603, ...</td>\n",
       "      <td>[0.4621147, -0.32601243, 0.14875361, 0.0588712...</td>\n",
       "      <td>[-0.17904884, -0.30857, -0.08816792, 0.2628701...</td>\n",
       "      <td>[0.3672084, -0.39905167, 0.45736057, -0.170849...</td>\n",
       "      <td>[0.15963465, 0.23201807, 0.07888236, -0.746187...</td>\n",
       "      <td>[-0.27435634, -0.34182993, 0.42234477, -0.0363...</td>\n",
       "      <td>[-0.8862241, -0.77771133, -0.3682679, -0.15156...</td>\n",
       "      <td>[-0.24318284, -1.1066128, -0.28148663, -0.1359...</td>\n",
       "      <td>[0.3766895, 0.3671902, 0.2224524, -0.2929603, ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>[-0.01027973, -0.5219526, 0.10329938, -0.09050...</td>\n",
       "      <td>[-0.45395166, -0.35621172, -0.22584479, -0.092...</td>\n",
       "      <td>[-0.2773134, -0.9491745, 0.41337347, -0.102893...</td>\n",
       "      <td>[0.08987117, 0.77899325, 0.32785872, -0.414579...</td>\n",
       "      <td>[-0.18151264, -0.15535061, 0.33438784, 0.08275...</td>\n",
       "      <td>[-0.4375131, -0.1936216, 0.03947416, -0.094286...</td>\n",
       "      <td>[0.48783925, 0.17583346, -0.14498003, -0.67800...</td>\n",
       "      <td>[0.42487782, 0.26128945, 0.3736353, -0.1579544...</td>\n",
       "      <td>[-0.6800742, -0.19899178, 0.8965055, -0.387440...</td>\n",
       "      <td>[-0.22371355, -0.27262712, 0.19397196, 0.47106...</td>\n",
       "      <td>[-0.50069714, -0.51396066, -0.0030787587, 0.34...</td>\n",
       "      <td>[0.30327797, 0.552616, 0.021566426, -0.6280945...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[0.51798856, -0.014191009, -0.6117699, 0.43670...</td>\n",
       "      <td>[-0.25571728, -0.30425158, -0.52816045, -0.173...</td>\n",
       "      <td>[0.16311847, 0.115701005, 0.2898489, -0.367323...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              city#1  \\\n",
       "0  [-0.17081124, -0.64723957, 0.22311965, 0.11456...   \n",
       "1  [0.28466508, -0.6870391, 0.034863576, 0.143425...   \n",
       "2  [0.008446187, -0.89799815, 0.3153159, -0.01146...   \n",
       "3  [0.4621147, -0.32601243, 0.14875361, 0.0588712...   \n",
       "4  [-0.01027973, -0.5219526, 0.10329938, -0.09050...   \n",
       "\n",
       "                                           country#1  \\\n",
       "0  [-0.1733922, -0.29064077, 0.08646101, 0.469208...   \n",
       "1  [-0.45580024, -0.64020044, -0.94991213, 1.1212...   \n",
       "2  [-0.08009297, -0.70065546, 0.8756641, -0.21270...   \n",
       "3  [-0.17904884, -0.30857, -0.08816792, 0.2628701...   \n",
       "4  [-0.45395166, -0.35621172, -0.22584479, -0.092...   \n",
       "\n",
       "                                      company_name#1  \\\n",
       "0  [-0.2278746, -0.33084768, -0.6021811, 0.147618...   \n",
       "1  [-0.077516824, -0.10479821, 0.33265123, -0.063...   \n",
       "2  [-0.05640249, -0.50657225, 0.198513, 0.1292749...   \n",
       "3  [0.3672084, -0.39905167, 0.45736057, -0.170849...   \n",
       "4  [-0.2773134, -0.9491745, 0.41337347, -0.102893...   \n",
       "\n",
       "                                             phone#1  \\\n",
       "0  [0.1663901, 0.4832298, 0.31022692, -0.52979314...   \n",
       "1  [0.10064737, 0.48980132, 0.121493064, -0.52424...   \n",
       "2  [0.22372286, 0.7134844, 0.27984577, -0.703583,...   \n",
       "3  [0.15963465, 0.23201807, 0.07888236, -0.746187...   \n",
       "4  [0.08987117, 0.77899325, 0.32785872, -0.414579...   \n",
       "\n",
       "                                    street_address#1  \\\n",
       "0  [0.13965015, -0.3673879, 0.4392662, -0.1811918...   \n",
       "1  [-0.10337315, -0.25847742, 0.5891691, -0.21212...   \n",
       "2  [-0.26451614, -0.3932885, 0.4730599, -0.185341...   \n",
       "3  [-0.27435634, -0.34182993, 0.42234477, -0.0363...   \n",
       "4  [-0.18151264, -0.15535061, 0.33438784, 0.08275...   \n",
       "\n",
       "                                           website#1  \\\n",
       "0  [0.8729764, -0.21711689, 0.14693439, 0.1163865...   \n",
       "1  [0.01968214, 0.6168068, -0.90221906, 0.0048712...   \n",
       "2  [0.19262332, -0.031438783, -0.19407359, 0.1009...   \n",
       "3  [-0.8862241, -0.77771133, -0.3682679, -0.15156...   \n",
       "4  [-0.4375131, -0.1936216, 0.03947416, -0.094286...   \n",
       "\n",
       "                                      linkedin_url#1  \\\n",
       "0  [0.30312464, -0.3625717, 0.23163374, 0.4615468...   \n",
       "1  [-0.456138, -0.53931075, -0.30390802, 0.474316...   \n",
       "2  [-0.28189784, -0.4292636, 0.48723325, -0.07348...   \n",
       "3  [-0.24318284, -1.1066128, -0.28148663, -0.1359...   \n",
       "4  [0.48783925, 0.17583346, -0.14498003, -0.67800...   \n",
       "\n",
       "                                      facebook_url#1  \\\n",
       "0  [0.37542623, 0.53012043, 0.40159354, -0.314179...   \n",
       "1  [0.21144052, 0.20298834, 0.21815896, -0.486129...   \n",
       "2  [0.5606265, 0.4351116, 0.43130416, -0.25141925...   \n",
       "3  [0.3766895, 0.3671902, 0.2224524, -0.2929603, ...   \n",
       "4  [0.42487782, 0.26128945, 0.3736353, -0.1579544...   \n",
       "\n",
       "                                              city#2  \\\n",
       "0  [-0.17081124, -0.64723957, 0.22311965, 0.11456...   \n",
       "1  [0.28466508, -0.6870391, 0.034863576, 0.143425...   \n",
       "2  [-0.21692994, -0.59062845, -0.63720477, 0.0989...   \n",
       "3  [0.4621147, -0.32601243, 0.14875361, 0.0588712...   \n",
       "4  [-0.6800742, -0.19899178, 0.8965055, -0.387440...   \n",
       "\n",
       "                                           country#2  \\\n",
       "0  [-0.1733922, -0.29064077, 0.08646101, 0.469208...   \n",
       "1  [-0.45580024, -0.64020044, -0.94991213, 1.1212...   \n",
       "2  [-0.2784196, -0.5214716, -0.2022338, 0.1992558...   \n",
       "3  [-0.17904884, -0.30857, -0.08816792, 0.2628701...   \n",
       "4  [-0.22371355, -0.27262712, 0.19397196, 0.47106...   \n",
       "\n",
       "                                      company_name#2  \\\n",
       "0  [-0.2278746, -0.33084768, -0.6021811, 0.147618...   \n",
       "1  [-0.3959472, -0.06375926, 0.039568715, 0.15786...   \n",
       "2  [-0.05473563, -0.11676584, 0.12873732, -0.0655...   \n",
       "3  [0.3672084, -0.39905167, 0.45736057, -0.170849...   \n",
       "4  [-0.50069714, -0.51396066, -0.0030787587, 0.34...   \n",
       "\n",
       "                                             phone#2  \\\n",
       "0  [0.1663901, 0.4832298, 0.31022692, -0.52979314...   \n",
       "1  [0.10064737, 0.48980132, 0.121493064, -0.52424...   \n",
       "2  [0.12499511, 0.7256371, 0.2397831, -0.56979215...   \n",
       "3  [0.15963465, 0.23201807, 0.07888236, -0.746187...   \n",
       "4  [0.30327797, 0.552616, 0.021566426, -0.6280945...   \n",
       "\n",
       "                                    street_address#2  \\\n",
       "0  [0.054061692, -0.09831729, 0.62615895, 0.04232...   \n",
       "1  [-0.10337315, -0.25847742, 0.5891691, -0.21212...   \n",
       "2  [-0.23111165, -0.4517533, 0.1346171, -0.219431...   \n",
       "3  [-0.27435634, -0.34182993, 0.42234477, -0.0363...   \n",
       "4  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "\n",
       "                                           website#2  \\\n",
       "0  [0.8729764, -0.21711689, 0.14693439, 0.1163865...   \n",
       "1  [0.01968214, 0.6168068, -0.90221906, 0.0048712...   \n",
       "2  [0.0700393, -0.6345332, 0.08723115, 0.00782650...   \n",
       "3  [-0.8862241, -0.77771133, -0.3682679, -0.15156...   \n",
       "4  [0.51798856, -0.014191009, -0.6117699, 0.43670...   \n",
       "\n",
       "                                      linkedin_url#2  \\\n",
       "0  [-0.35564625, 0.0014702752, -0.2596981, -0.141...   \n",
       "1  [-0.6107266, -1.229, 0.017428547, -0.58206856,...   \n",
       "2  [0.3146601, -0.14218765, 0.77169657, -0.107591...   \n",
       "3  [-0.24318284, -1.1066128, -0.28148663, -0.1359...   \n",
       "4  [-0.25571728, -0.30425158, -0.52816045, -0.173...   \n",
       "\n",
       "                                      facebook_url#2  match  \n",
       "0  [0.37542623, 0.53012043, 0.40159354, -0.314179...      1  \n",
       "1  [0.21144052, 0.20298834, 0.21815896, -0.486129...      1  \n",
       "2  [0.533215, 0.30011246, 0.2952332, -0.20062599,...      0  \n",
       "3  [0.3766895, 0.3671902, 0.2224524, -0.2929603, ...      1  \n",
       "4  [0.16311847, 0.115701005, 0.2898489, -0.367323...      0  "
      ]
     },
     "execution_count": 189,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vectorized_companies_dataframe.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predictors: ['city#1', 'country#1', 'company_name#1', 'phone#1', 'street_address#1', 'website#1', 'linkedin_url#1', 'facebook_url#1', 'city#2', 'country#2', 'company_name#2', 'phone#2', 'street_address#2', 'website#2', 'linkedin_url#2', 'facebook_url#2']\n",
      "Predicted: match\n",
      "X (10000, 16)\n",
      "Y (10000,)\n",
      "Train (7200, 16) (7200,)\n",
      "Test (2000, 16) (2000,)\n",
      "Val (800, 16) (800,)\n",
      "Train: (7200, 16, 96) (7200,)\n",
      "Test: (2000, 16, 96) (2000,)\n",
      "Val: (800, 16, 96) (800,)\n",
      "Matches: [3616, 1004, 402] [7200, 2000, 800]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "x_train, y_train, x_test, y_test, x_val, y_val = split_dataframe(\n",
    "    dataset=vectorized_companies_dataframe,\n",
    "    test_size=0.2,\n",
    "    val_size=0.1\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input: (7200, 16, 96)\n",
      "Model: \"model_7\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_8 (InputLayer)         [(None, 16, 96)]          0         \n",
      "_________________________________________________________________\n",
      "flatten_7 (Flatten)          (None, 1536)              0         \n",
      "_________________________________________________________________\n",
      "dense_35 (Dense)             (None, 4096)              6295552   \n",
      "_________________________________________________________________\n",
      "dropout_28 (Dropout)         (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dense_36 (Dense)             (None, 2048)              8390656   \n",
      "_________________________________________________________________\n",
      "dropout_29 (Dropout)         (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_37 (Dense)             (None, 1024)              2098176   \n",
      "_________________________________________________________________\n",
      "dropout_31 (Dropout)         (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_39 (Dense)             (None, 1)                 1025      \n",
      "=================================================================\n",
      "Total params: 16,785,409\n",
      "Trainable params: 16,785,409\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Epoch 1/11\n",
      "72/72 [==============================] - 45s 574ms/step - loss: 0.7734 - binary_accuracy: 0.5574 - val_loss: 0.5774 - val_binary_accuracy: 0.7905\n",
      "Epoch 2/11\n",
      "72/72 [==============================] - 28s 386ms/step - loss: 0.5416 - binary_accuracy: 0.7278 - val_loss: 0.4057 - val_binary_accuracy: 0.8125\n",
      "Epoch 3/11\n",
      "72/72 [==============================] - 30s 415ms/step - loss: 0.3709 - binary_accuracy: 0.8380 - val_loss: 0.2315 - val_binary_accuracy: 0.9345\n",
      "Epoch 4/11\n",
      "72/72 [==============================] - 28s 392ms/step - loss: 0.2179 - binary_accuracy: 0.9129 - val_loss: 0.1987 - val_binary_accuracy: 0.9115\n",
      "Epoch 5/11\n",
      "72/72 [==============================] - 35s 494ms/step - loss: 0.1805 - binary_accuracy: 0.9334 - val_loss: 0.1206 - val_binary_accuracy: 0.9585\n",
      "Epoch 6/11\n",
      "72/72 [==============================] - 33s 451ms/step - loss: 0.0934 - binary_accuracy: 0.9687 - val_loss: 0.9856 - val_binary_accuracy: 0.5650\n",
      "Epoch 7/11\n",
      "72/72 [==============================] - 38s 528ms/step - loss: 0.1293 - binary_accuracy: 0.9575 - val_loss: 0.1773 - val_binary_accuracy: 0.9360\n",
      "Epoch 8/11\n",
      "72/72 [==============================] - 23s 322ms/step - loss: 0.1394 - binary_accuracy: 0.9744 - val_loss: 0.0899 - val_binary_accuracy: 0.9640\n",
      "Epoch 9/11\n",
      "72/72 [==============================] - 28s 386ms/step - loss: 0.0474 - binary_accuracy: 0.9923 - val_loss: 0.0788 - val_binary_accuracy: 0.9675\n",
      "Epoch 10/11\n",
      "72/72 [==============================] - 32s 438ms/step - loss: 0.0517 - binary_accuracy: 0.9947 - val_loss: 0.0875 - val_binary_accuracy: 0.9700\n",
      "Epoch 11/11\n",
      "72/72 [==============================] - 30s 420ms/step - loss: 0.0440 - binary_accuracy: 0.9949 - val_loss: 0.0925 - val_binary_accuracy: 0.9660\n",
      "{'loss': [0.688974142074585, 0.5073610544204712, 0.34385862946510315, 0.2104153037071228, 0.18272976577281952, 0.12562739849090576, 0.10655010491609573, 0.15416139364242554, 0.07874878495931625, 0.10132322460412979, 0.10475065559148788], 'binary_accuracy': [0.6038888692855835, 0.7511110901832581, 0.8552777767181396, 0.9127777814865112, 0.9337499737739563, 0.9581944346427917, 0.9708333611488342, 0.971666693687439, 0.9858333468437195, 0.9895833134651184, 0.9877777695655823], 'val_loss': [0.5773686766624451, 0.40574324131011963, 0.23154781758785248, 0.19872945547103882, 0.12060342729091644, 0.9856425523757935, 0.1772746592760086, 0.08985311537981033, 0.07883580029010773, 0.08746380358934402, 0.09246744215488434], 'val_binary_accuracy': [0.7904999852180481, 0.8125, 0.934499979019165, 0.9114999771118164, 0.9585000276565552, 0.5649999976158142, 0.9359999895095825, 0.9639999866485596, 0.9674999713897705, 0.9700000286102295, 0.9660000205039978]}\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8rg+JYAAAACXBIWXMAAAsTAAALEwEAmpwYAAA6wElEQVR4nO3dd3yV5dnA8d91MskkG5IAQWbCUGSIo+4BuLUqCFZ9rVhbV61UtNbW1rZ2+Wr7qhUVFy4cbR0ouLVVZKnIShgyQhgnAZKQPe73j+ccCDFAcvI8Z17fz+d8zsnznHOf6zDOlXs81y3GGJRSSkUuV6ADUEopFViaCJRSKsJpIlBKqQiniUAppSKcJgKllIpwmgiUUirCaSJQqhNEpEBEjIhEd+K5V4nIf7rbjlL+oolAhR0R2SgijSKS2e74l54v4YIAhaZUUNJEoMLVt8AU7w8iMgJICFw4SgUvTQQqXD0L/KDNz1cCz7R9goikisgzIuIWkU0icpeIuDznokTkLyJSLiIbgLM7eO0TIrJNRLaKyL0iEtXVIEUkV0ReF5FdIrJORK5tc26ciCwRkSoR2SEi93uOx4vIHBGpEJE9IrJYRHK6+t5KeWkiUOFqIZAiIoWeL+jJwJx2z/k7kAocAZyElTiu9py7FjgHGAWMAb7f7rVPAc3AQM9zzgR+6EOcLwKlQK7nPX4vIqd6zj0IPGiMSQEGAHM9x6/0xN0HyAB+BNT58N5KAZoIVHjz9grOAFYDW70n2iSHO4wx1caYjcBfgSs8T7kUeMAYs8UYswv4Q5vX5gCTgFuMMTXGmJ3A/3ra6zQR6QMcD9xujKk3xnwFPM7+nkwTMFBEMo0xe40xC9sczwAGGmNajDFLjTFVXXlvpdrSRKDC2bPA5cBVtBsWAjKBGGBTm2ObgDzP41xgS7tzXv08r93mGZrZAzwKZHcxvlxglzGm+iAxXAMMBtZ4hn/OafO55gMvikiZiPxJRGK6+N5K7aOJQIUtY8wmrEnjScBr7U6XY/1m3a/Nsb7s7zVswxp6aXvOawvQAGQaY3p6binGmGFdDLEMSBeR5I5iMMasNcZMwUowfwReEZFEY0yTMeYeY0wRcBzWENYPUMpHmghUuLsGONUYU9P2oDGmBWvM/Xcikiwi/YBb2T+PMBe4SUTyRSQNmNnmtduABcBfRSRFRFwiMkBETupKYMaYLcBnwB88E8AjPfHOARCRaSKSZYxpBfZ4XtYqIqeIyAjP8FYVVkJr7cp7K9WWJgIV1owx640xSw5y+kagBtgA/Ad4HpjtOfcY1vDL18Ayvtuj+AEQC6wCdgOvAL19CHEKUIDVO/gn8CtjzHuecxOAlSKyF2vieLIxpg7o5Xm/Kqy5j4+xhouU8onoxjRKKRXZtEeglFIRThOBUkpFOE0ESikV4TQRKKVUhAu5UriZmZmmoKAg0GEopVRIWbp0abkxJqujcyGXCAoKCliy5GCrAZVSSnVERDYd7JwODSmlVITTRKCUUhHOsUQgIrNFZKeIrDjIeRGRv3lqsC8XkaOdikUppdTBOTlH8BTwf3y36qPXRGCQ53YM8IjnvsuampooLS2lvr7el5eHjPj4ePLz84mJ0UKTSin7OJYIjDGfHGZv2POBZ4xV42KhiPQUkd6egl5dUlpaSnJyMgUFBYiIryEHNWMMFRUVlJaW0r9//0CHo5QKI4GcI8jjwHrvpeyvw34AEZnu2bJvidvt/s75+vp6MjIywjYJAIgIGRkZYd/rUUr5X0hMFhtjZhljxhhjxmRldbgMNqyTgFckfEallP8FMhFs5cCNP/Jps5WgUiGluRGWPg2tLYGORKkuC2QieB34gWf10Hig0pf5gWCwZ88eHn744S6/btKkSezZs8f+gJT/lbwNb9wEGz4MdCRKdZmTy0dfAD4HhohIqYhcIyI/EpEfeZ4yD2tDkHVYm4D82KlYnHawRNDc3HzI182bN4+ePXs6FJXyq51rrPsdqwIbh1I+cHLV0JTDnDfAT5x6f3+aOXMm69ev56ijjiImJob4+HjS0tJYs2YNJSUlXHDBBWzZsoX6+npuvvlmpk+fDuwvl7F3714mTpzICSecwGeffUZeXh7//ve/6dGjR4A/meq08mLrfufqwMahlA9CrtbQ4dzzxkpWlVXZ2mZRbgq/Ovfg+5Lfd999rFixgq+++oqPPvqIs88+mxUrVuxb5jl79mzS09Opq6tj7NixXHzxxWRkZBzQxtq1a3nhhRd47LHHuPTSS3n11VeZNm2arZ9DOchdYt3v1B6BCj1hlwiCwbhx4w5Y6/+3v/2Nf/7znwBs2bKFtWvXficR9O/fn6OOOgqA0aNHs3HjRn+Fq7qrtQXKPYnAXWz97IoKbExKdUHYJYJD/ebuL4mJifsef/TRR7z33nt8/vnnJCQkcPLJJ3d4LUBcXNy+x1FRUdTV1fklVmWDPZugpQH6jIctC2H3RsgYEOiolOq0kLiOINglJydTXV3d4bnKykrS0tJISEhgzZo1LFy40M/RKcd5h4WGXWjd6/CQCjGaCGyQkZHB8ccfz/Dhw5kxY8YB5yZMmEBzczOFhYXMnDmT8ePHByhK5Ri3Z8VQ0XnWvU4YqxATdkNDgfL88893eDwuLo633367w3PeeYDMzExWrNhfpPW2226zPT7loPISSMqBlFxIK9AegQo52iNQqrvcxZA1xHqcPUyvJVAhRxOBUt1hjJUIMr2JoBAq1kFzQ2DjUqoLNBEo1R3V26Cxuk2PoBBMC5SvDWxcSnWBJgKlusPtuaJ4XyIosu51wliFEE0ESnWHNxF4h4YyBoIrWieMVUjRRKBUd5QXQ3wqJGVbP0fHQuZgTQQqpGgisIGvZagBHnjgAWpra22OSPmNuwSyhkLbTYOyCzURqJCiicAGmggimHuN1QNoK7sQ9myGho6vNlcq2OgFZTZoW4b6jDPOIDs7m7lz59LQ0MCFF17IPffcQ01NDZdeeimlpaW0tLTwy1/+kh07dlBWVsYpp5xCZmYmH36om5qElNpdUFu+f6LYyzth7C6G/DH+j0upLgq/RPD2TNj+jb1t9hoBE+876Om2ZagXLFjAK6+8wqJFizDGcN555/HJJ5/gdrvJzc3lrbfeAqwaRKmpqdx///18+OGHZGZm2huzct6+FUNDDzzuTQQ7VmoiUCFBh4ZstmDBAhYsWMCoUaM4+uijWbNmDWvXrmXEiBG8++673H777Xz66aekpqYGOlTVXd7NaNoPDfXsBzEJuoRUhYzw6xEc4jd3fzDGcMcdd3Ddddd959yyZcuYN28ed911F6eddhp33313ACJUtnEXW1/4qX0OPO5yWb0EnTBWIUJ7BDZoW4b6rLPOYvbs2ezduxeArVu3snPnTsrKykhISGDatGnMmDGDZcuWfee1KsS4iyFzkPXF3152kfYIVMgIvx5BALQtQz1x4kQuv/xyjj32WACSkpKYM2cO69atY8aMGbhcLmJiYnjkkUcAmD59OhMmTCA3N1cni0NNeQn0Pbbjc9mF8NUcqCmHRJ3/UcFNE4FN2pehvvnmmw/4ecCAAZx11lnfed2NN97IjTfe6GhsygENe6FyC2Rd2fH5HG+piVXQ/0T/xaWUD3RoSClfePcobr9iyEtrDqkQoolAKV94E0HmkI7PJ+VAjzSdMFYhIWwSgTEm0CE4LhI+Y8hwr7GKy6X37/i8iE4Yq5ARFokgPj6eioqKsP6iNMZQUVFBfHx8oENRYNUYyhgIUTEHf052oZUIwvjfpQoPYTFZnJ+fT2lpKW63O9ChOCo+Pp78/PxAh6HAupjMOw9wMNlF0FAFlaXQs8+hn6tUAIVFIoiJiaF//4N00ZWyW3MD7NoAwy489PPaThhrIlBBLCyGhpTyq4r1YFoPvmLIK9tzXieMVZDTRKBUVx2sxlB7PdIgOVcnjFXQ00SgVFe5iwGxykscTk4R7FzpeEhKdYcmAqW6yl0Maf0gpsfhn5tdaK0waml2Pi6lfKSJQKmuKi85+IVk7WUXQUsD7P7W2ZiU6gZNBEp1RWsLlK+FrMPMD3hlF1r3OmGsgpgmAqW6YvdG6zf8zvYIMocAohPGKqhpIlCqKw5XbK692ARIP8LatlKpIOVoIhCRCSJSLCLrRGRmB+f7isiHIvKliCwXkUlOxqNUt7nXWPedHRqC/aUmlApSjiUCEYkCHgImAkXAFBFpf03+XcBcY8woYDLwsFPxKGULdwkk9YL4Luw5nV0Eu9ZDU71zcSnVDU72CMYB64wxG4wxjcCLwPntnmOAFM/jVKDMwXiU6r7yYsjq5PyAV3ahdSWyd1hJqSDjZCLIA7a0+bnUc6ytXwPTRKQUmAd0uFWXiEwXkSUisiTcC8upIGaM1SPoaiLIGWbd68ohFaQCPVk8BXjKGJMPTAKeFZHvxGSMmWWMGWOMGZOVleX3IJUCoKoMGqsPX1qivfQjICpWE4EKWk4mgq1A25KL+Z5jbV0DzAUwxnwOxAO607cKTt4aQ51dMeQVFWMlD50wVkHKyUSwGBgkIv1FJBZrMvj1ds/ZDJwGICKFWIlAx35UcHJ7E0EXh4ZAVw6poOZYIjDGNAM3APOB1Virg1aKyG9E5DzP034GXCsiXwMvAFeZcN5mTIU2dzHE94REH4YnswuhcgvUV9kellLd5ejGNMaYeViTwG2P3d3m8SrgeCdjUMo25SXWsJBI11+b7Z0wXg19j7E3LqW6KdCTxUqFDvearl1I1pbWHFJBTBOBUp1RUwG1FZ2vMdReah+ITdJ5AhWUNBEo1Rm+rhjycrms12qPQAUhTQRKdYYvNYbayymyEoGuh1BBRhOBUp3hLoGYBEjJ972N7CJreKlGV0ir4KKJQKnOKC+2LgpzdeO/jE4YqyCliUCpznD7UGyuvWxP8V2dMFZBRhOBUofTUA1VW7teY6i9xCxIyNAegQo6mgiUOpyu7kp2MCJWr2CHJgIVXDQRKHU43akx1F52kbUCqbW1+20pZRNNBEodjrsYXDGQ1r/7bWUXQuNeq+6QUkFCE4FSh1NeAhkDIcqG0lw6YayCkCYCpQ7HXdy9C8na2reEdKU97SllA00ESh1KUz3s/tb3GkPtxadYdYe0R6CCiCYCpQ5l13pr43k7Joq9dJMaFWQ0ESh1KHauGPLKLrTmHVqa7GtTqW6IqETQ2KxL9lQXuYsBsSaL7ZJdBC2NsGuDfW0q1Q0RkwjmLtnChAc/obJOfwtTXVBeDGkFENPDvja9K4d26ISxCg4RkwgGZCWxuaKWn839mtZWLQOsOsldYu+wEFilKsSl8wQqaERMIhjdL41fnF3Ie6t3MOtT7ZKrTmhphoq13a8x1F5MPKQP0JpDKmhETCIAuOq4As4e2Zs/vbOGhRsqAh2OCnZ7Nllj+d2tMdQRXTmkgkhEJQIR4Y8Xj6QgM5Ebnv+SnVX1gQ5JBTMnVgx55QyzJosba+1vW6kuiqhEAJAUF80/po2mpqGZG174kuYWXUmkDsK7PWXmIPvbzi4EzP69kJUKoIhLBACDc5L5w0UjWPTtLv48X/8jqoMoL4HkXIhPtb9trTmkgkhEJgKAC0blMW18Xx79ZAPzV24PdDgqGNlZY6i9tP4QFacTxiooRGwiAPjlOUUcmZ/KbXO/ZmN5TaDDUcHEGKtHYFeNofaioq0koz0CFQQiOhHERUfx0NSjcbmE659bRn1TS6BDUsGiaqu1b4ATE8Ve2cN0tzIVFCI6EQDkpyXwwGVHsXpbFXf/e0Wgw1HBwskVQ17ZhVBdBnW7nXsPpToh4hMBwClDs7nx1IHMXVLK3MW6c5RifyJwamgI2kwYr3HuPZTqBE0EHrecPpgTBmbyy3+vYGVZZaDDUYFWXgw90iEx07n32LdJjQ4PqcDSROAR5RIenHwUaQmxXD9nmRani3TeGkMizr1Haj7EpWgiUAGniaCNjKQ4Hpp6NGV76rjt5a8xRovTRSz3GvtrDLUnoqUmVFDQRNDO6H5p3DmpkHdX7WDWJ1qcLiLVlEPdLmdqDLWXXWj1CPSXDhVAmgg6cPXxBZw9ojd/ml/MF1qcLvLsWzHkcI8ArAnjut2wd4fz76XUQTiaCERkgogUi8g6EZl5kOdcKiKrRGSliDzvZDydJSLcd/EI+qUncMMLWpwu4uyrMeTgiiEvnTBWQcCxRCAiUcBDwESgCJgiIkXtnjMIuAM43hgzDLjFqXi6Kjk+hkemjWZvvRanizjlJRCTaE3mOm3fbmWaCFTgONkjGAesM8ZsMMY0Ai8C57d7zrXAQ8aY3QDGmJ0OxtNlQ3ol8/uLhlvF6RZocbqI4a0x5OSKIa/ETEjM1gljFVBOJoI8oO3VWaWeY20NBgaLyH9FZKGITOioIRGZLiJLRGSJ2+12KNyOXTgqn6nH9OXRjzewQIvTRQZ3sX+Ghby8E8ZKBUigJ4ujgUHAycAU4DER6dn+ScaYWcaYMcaYMVlZWf6NELj73CJG5qfys5e/ZlOFFqcLa/VVVtkHf0wUe2UXWfMSrTr8qALDyUSwFejT5ud8z7G2SoHXjTFNxphvgRKsxBBU4qKjeOjyo3GJcP0cLU4X1srXWvf+WDrqlV0ITbWwZ6P/3lOpNpxMBIuBQSLSX0RigcnA6+2e8y+s3gAikok1VBSUi/f7pCfwv5cdyaptVfzq3ysDHY5yij9XDHnlDLPudZ5ABYhjicAY0wzcAMwHVgNzjTErReQ3InKe52nzgQoRWQV8CMwwxjizcL98Lbx7d7cu3Dl1aA43nDKQl5ZsYe4SLU4XlsqLISoW0gr8957eCqc6T6ACJNrJxo0x84B57Y7d3eaxAW713Jy1dgH890GrtsuJt/nczE/PGMyXW3bzy3+tYFhuCsNyHdjGUAWOuwQyBlobx/hLXDL07Ks9AhUwgZ4s9p/xP4YRl8AH90Lx2z43YxWnG0VaQiw/fk6L04Udf9QY6kh2kSYCFTCdSgQikigiLs/jwSJynojEOBuazUTgvL9D75Hw6rX7ywj4IDMpjoemjmLr7jpmaHG68NFUD3s2ObsZzcFkF1kXsjU3+v+9VcTrbI/gEyBeRPKABcAVwFNOBeWYmB4w+Xnr/oUp3doZanS/dO6YVMiCVTt47NOgnN9WXVWxDkxr4BJBa7MVg1J+1tlEIMaYWuAi4GFjzCXAMOfCclBqPlz2LOzZDK9cA62+LwX9n+MLmDSiF398R4vThYVArBjy0ppDKoA6nQhE5FhgKvCW51iUMyH5Qd/xcPZfYP378N6vfG5GRPjjxSP3F6er1uJ0Ia28BMRlTRb7W+YgkCidJ1AB0dlEcAtWcbh/epaAHoG13DN0jb4Kxl4Ln/0dvn7J52aS42N4eNrRVNc3cePzWpwupLmLrWWjMfH+f+/oOCsBaY9ABUCnEoEx5mNjzHnGmD96Jo3LjTE3ORyb8yb8AQq+B6/fCFuX+tzM0F4p/P7CEXzx7S7+sqDExgCVX5WXBGZYyCunSBOBCojOrhp6XkRSRCQRWAGsEpEZzobmB1ExcMnTkJQDL06Fat+Lyl10dD6XH9OXf3y8nndX6SYjIael2bro0J81htrLLoLdG6FR61kp/+rs0FCRMaYKuAB4G+iPtXIo9CVmwJTnob4SXroCmht8buruc4oYkZfKrXO/YnNFrY1BKsft3gitTf6tMdSed8LYO2mtlJ90NhHEeK4buABPkTggfBbP9xoBFzwMpYvgrZ/5XIYiPiaKh6d6itM9t1SL04WScs91JYEcGvJuUqMTxsrPOpsIHgU2AonAJyLSD6hyKqiAGHYhnDgDvnwWFs3yuRlvcbqVZVX8+nUtThcy9i0dDWDx27QCiO6hu5Upv+vsZPHfjDF5xphJxrIJOMXh2Pzv5DthyCR45w7Y8LHPzZw6NIefnDKAFxdv4WUtThca3CWQkgfxKYGLwRVlXcymE8bKzzo7WZwqIvd7dwkTkb9i9Q7Ci8sFFz5q/Vb48pXWuLGPbj1jCMcNyOCuf61gVVl4dZ7CUnlxYGoMtac1h1QAdHZoaDZQDVzquVUBTzoVVEDFp1hlKIyBFy6Hhr0+NRPlEv42ZRQ9E2L48XNLqarX4nRBq7XV6hEEorREe9mFsHc71O4KdCQqgnQ2EQwwxvzKsxH9BmPMPcARTgYWUBkD4JInwb0a/nmdz1sIZibF8dDlR1O6u46bXtCLzYJW1VZoqgmSROCdMNbhIeU/nU0EdSJygvcHETkeqHMmpCAx4FQ4815Y8yZ88mefmxlTkM5vzh/OR8Vufvum/ucOSsGwYsgrR1cOKf/r7O4bPwKeERHvLiy7gSudCSmIjP8xbP8GPvq9tZ1g4Tk+NXP5MX3ZWFHDrE82UJCZyNXH97c5UNUt3pLkwdAjSO4N8anaI1B+1dlVQ18bY44ERgIjjTGjgFMdjSwYiMA5D0DeaGuIqBvL+m6fMJQzi3L47ZureH+1XnkcVNzFkJABiZmBjsT6N6cTxsrPurRDmTGmynOFMfhje8lgEBMPlz0HsUnwwmSfJ/GiXMIDk49iWG4qN77wJSvLKm0OVPks0DWG2ssutHoEuuGR8pPubFUptkUR7FJ6w2VzoHobvHyVVZfGBwmx0Txx5RhSe8RwzVNL2F6pZasDzhjrYrJA1hhqL7vIKnlSVRboSFSE6E4iiKxfV/qMtYaJvv0YFtzlczPZKfHMvmos1fVNXPP0YmoafEsqyiY15dZOdYGsMdSelppQfnbIRCAi1SJS1cGtGsj1U4zBY9RUawL5i0fgyzk+N1PYO4X/u/xoVm+r4uYXv6KlNbJyalDZt2IomHoEuluZ8q9DJgJjTLIxJqWDW7IxprMrjsLLGb+F/ifBmz+FLYt9buaUodn86txhvLd6B7+fp7/5BYy3xlAwrBjySkiHpF7aI1B+052hocgUFQ2XPAUpufDSNKja5nNTVx5XwFXHFfDEf77l2YWb7ItRdZ67xFoIkJIX6EgO5J0wVsoPNBH4IiEdJr8AjXvhpanQ5Puk7y/PKeLUodn8+vWVfFzitjFI1SneGkMSZGsfcoZZvZVWLWWunKeJwFc5RVaBuq1L4Y2bfV7q561JNDgnmZ88t4w127VAnV+5i4NrWMgruxCa67tV+FCpztJE0B2F51ilq5e/CAsf9rmZpLhoZl81hsS4KK55agk7q3VZqV/UV1pLgoM1EYAODym/0ETQXSfOgMJzrSWl6z/wuZneqT144sqx7Kpp5Nqnl1DXqEMCjitfa90H08VkXllDAdEJY+UXmgi6y+WCC/4BWYXw8tVQsd7npobnpfLg5KNYvrWSW+d+RasuK3VWMK4Y8opNtHYs26G73CnnaSKwQ1wSTHnemnB8YQrU+z7Of+awXvxiUiFvr9jOn+YX2xik+g53MUTFQc9+gY6kY1pzSPmJJgK7pBXAJU9Dxbpu7WEAcM0J/Zl6TF/+8fF6Xly02b4Y1YHKSyBjoLUkOBhlF1r/npobAh2JCnOaCOx0xEkw4Q9QPM8qXe0jEeGe84Zx4uAs7vrXCv67rtzGINU+wVZjqL3sQjAt++cylHKIJgK7jZsOo66wNrNZ+U+fm4mOcvHQ5aMYkJXEj+YsZd3OahuDVDTVwe5NwVVjqD2tOaT8RBOB3UTg7L9C/jj4l2djGx8lx8fwxFVjiIuO4uqnFlOxV4cIbFOxDjDBVWOovYyB4IqBnTphrJylicAJ0XFw2bMQ3xNeuBxqKnxuKj8tgcevHMPOqgamP7uU+iZdVmqLYNqV7GCiYyFzkPYIlOMcTQQiMkFEikVknYjMPMTzLhYRIyJjnIzHr5J7weQ5sHcHvHwltDT53NRRfXryv5cdxdJNu5nxynJdVmoHdzGIy/qtO5hpzSHlB44lAhGJAh4CJgJFwBQRKergecnAzcAXTsUSMHmj4by/w8ZPYf6d3Wpq0oje3D5hKG98XcYD75XYFGAEKy+GtP5W7y2YZRfCns3QoHNEyjlO9gjGAeuMMRuMMY3Ai8D5HTzvt8AfgfCsq3DkZXDsDbBoVrf2MAD40UlHcNmYPvztg3W8urTUpgAjVLDWGGpv34TxmsDGocKak4kgD9jS5udSz7F9RORooI8x5q1DNSQi00VkiYgscbtDsELn6ffs38OgdKnPzYgI9144nOMGZDDzteV8scH3uYeI1tJsXQEezBPFXvsSgQ4PKecEbLJYRFzA/cDPDvdcY8wsY8wYY8yYrKws54Ozm3cPg+Re1h4G1Tt8biomysUjU0fTNz2B6+Ys5dvyGvvijBS7v4XWpuBeOurVsx/EJOiEsXKUk4lgK9Cnzc/5nmNeycBw4CMR2QiMB14PqwnjthLSYfLz1v64L18JzY0+N5WaEMOTV43DJcLVTy5id43vbUWkfTWGQqBH4HJZCUt7BMpBTiaCxcAgEekvIrHAZOB170ljTKUxJtMYU2CMKQAWAucZY5Y4GFNg9RoB5/8fbP4c5t/Rrab6ZiTw2A9GU1ZZz3XPLqWhWZeVdpo7CPcpPhStOaQc5lgiMMY0AzcA84HVwFxjzEoR+Y2InOfU+wa9Ed+H426CxY/Dsme61dTofun8+fsjWbRxF3e8+g3Gx81xIk55CaTkQ1xyoCPpnJwiqNkJNVpqRDnD0Wpbxph5wLx2x+4+yHNPdjKWoHL6r60rjt/6mVW+us9Yn5s6/6g8NlXUcv+7JRRkJnLTaYPsizNcuYtDY1jIq+0mNf1PDGwsKizplcWB4IqC78+G5N4w94puTR4D3HjqQC46Oo/73y3h319tPfwLIllrq9UjCMbNaA5Gaw4ph2kiCBTv5HF9Jcz9Qbcmj0WEP1w0gnH905nxynKWbtplY6BhpqoUmmpD4xoCr6Qc6JGmE8bKMZoIAqnXcDj/IdiyEN65vVtNxUVH8ei00eT17MG1zyxlc0WtTUGGGbfnquxQSgQiVq9ghyYC5QxNBIE2/CI4/hZYMhuWPtWtptISY5l91VhajeHqpxZRWet7faOw5V06GkpDQ7B/5ZAuCFAO0EQQDE67GwacBm/dBlsWdaup/pmJPDptNJt31XL9c0tpbPZ9p7SwVF4MCZmQmBHoSLomuxAaq6FSS4so+2kiCAauKLj4cUjNg5eugKpt3WrumCMyuO+ikXy2voI7//mNVitty10SWsNCXjphrBykiSBYeCePG6o9k8fd24Tm4tH53HL6IF5ZWsqMV5bT3KI9A4yxhoZC5UKytrI95TB0wlg5QBNBMMkZBhc8DKWLYN6Mbjd382mDuPWMwby6rJSbX/xKh4lq3FC/JzRqDLXXIw1S8jQRKEc4ekGZ8sGwC2DbrfCf+yH3KBjzPz43JSLcdNogEmKjuPet1dQ3tfDQ1KOJj4myLdyQsm9XshDsEYBuUqMcoz2CYHTqXTDwDJj3c9i8sNvN/fB7R/C7C4fzQfFOrnl6MbWNzTYEGYJCdcWQV3ahNcfREqF/f8oxmgiCkSsKLn4MUvOt+YJuTh4DTD2mH3+95Eg+X1/BD55YRFV9BC4tLS+B2GRIyQ10JL7JLoKWBquMtlI20kQQrHqkeSaP91p7GHRz8hjgoqPzeejyo/m6dA9TH/si8spXe2sMiQQ6Et94aw7tWBnYOFTY0UQQzHKK4MJ/wNYlVoE6Gy4mmjiiN7OuGEPxjmomz1rIzurw3CG0Q+7i0B0WAs8kt+gSUmU7TQTBrug8+N5t8OWzsOQJW5o8ZWg2T101li27a5n86ELK9tTZ0m5Qq6+EvdtD8xoCr5gekH6EThgr22kiCAWn3AmDzoS3b4dNn9vS5HEDM3n2mnG49zZwyT8+Z1NFmG95GYo1hjqSXag9AmU7TQShwBUFFz1m7V879wdQaU+p6dH90nnh2vHUNjZz6aOfs25ntS3tBqV9K4ZCdOmoV3YR7FoPTRE0pKccp4kgVPToaU0eN9VaexjY9EUwPC+VF6cfS6uByx5dyMqySlvaDTrlxRAVB2kFgY6ke3KKwLRan0cpm2giCCXZQ+HCR2HrUtsmjwGG9Epm7nXHEhftYsqshXy5ebct7QYVdwlkDrJ6V6FMaw4pB2giCDWF58CJP4ev5lj7Htukf2Yic390LGmJsUx7/AsWbqiwre2gEKo1htpLPwKiYnXCWNlKE0EoOvkOGDwB3pkJG/9rW7P5aQm8fN2x5PbswZWzF/FR8U7b2g6opjrYszk0awy1FxVjJTTtESgbaSIIRS4XXDTLGu9++Upba9Rnp8Tz4vTxDMxO4tpnlvDOiu22tR0w5WsBE7o1htrLLtTdypStNBGEqvhUz+RxvXXlcZN91wJkJMXx/LXjGZGXyk+eX8a/v7JnlVLAeIvNhfLFZG1lF1l7L9eH6cS+8jtNBKEsa4jVMyj7Et681dZtDFN7xPDsNccwriCdW176ipcWb7atbb8rLwaJgowBgY7EHvsmjNcENg4VNjQRhLqhk+CkmfD187Bolq1NJ8ZF8+TVYzl5cBa3v/oNT/43RIuduYshvT9ExwU6Ent4aw7phLGyiSaCcHDS7TBkErxzB2z8j61Nx8dE8egVY5g4vBf3vLGKhz5cZ2v7fhHqNYbaS+0DsUk6Yaxso4kgHLhc1vUFGQNg7pWwZ4utzcdGu/j7lFFcOCqPP88v5s/z12BsHIZyVEuTdSVuqJeWaMvl0k1qlK00EYSL+BRr8rilEV6aauvkMUB0lIu/XnIkU8b15aEP1/ObN1eFRjLY9S20NodXIgDPyqGVts4LqciliSCcZA6yJo+3fQ1v3GL7l4TLJfz+wuH8z/H9efK/G7njtW9oaQ3yL6JwqTHUXnYR1O2y9mFWqps0EYSbIRPh5Dth+YvwxT9sb15E+OU5hdx46kBeXLyFW+d+RXNLq+3vYxtvTZ6wSwQ6Yazso5vXh6MTZ8D25TD/F9YXxhEn29q8iPCzM4fQIzaKP71TTH1TC3+bMoq46CCs4+MusSZX45ICHYm9vEtId6yy/e9XRR7tEYQjlwsueAQyBsIz58PfR8O/fwJfPgcV620bMvrxyQP59blFzF+5g+nPLKWuscWWdm0VLjWG2kvKhoRM7REoW2iPIFzFp8CVb1hDRJsXwpq34Ms51rnEbOg7Hvoea933GglRvv1TuOr4/iTERnP7a8u56slFPHHVWJLiguSfVWurVV6i4HuBjsQZukmNskmQ/I9VjkjOgeNvtm6trVCxFjZ9ZiWGzZ/D6tet58UkQp+x+xND3pguDaVcOrYP8bFR/PSlr7jiiS946qpxpCbEOPShuqByCzTXhU+Nofayi+Cr56y/W5d27pXvNBFECpfLWkKZNQTGXG0dqyrbnxQ2fw4f3QcYqxxD75HQ9zhPz2G8NRRxCOcdmUt8tIsbnv+SKY8t5NlrxpGRFOArecOtxlB72YXQuNdKeGn9Ah2NCmGOJgIRmQA8CEQBjxtj7mt3/lbgh0Az4Ab+xxizycmYVBspuTD8IusGVhGz0sWe5LAQljwBCx+yzqUP2N9j6HecVRdf5IDmzhzWi8evHMP0Z5dw2ayFXHfiEYzIT2VgVhLRUQH4jdW7YijcriHwyhlm3e9cpYlAdYtjiUBEooCHgDOAUmCxiLxujGk7u/UlMMYYUysi1wN/Ai5zKiZ1GPGpMPB06wbQ3Ghdk7D5cysxFM+zNsQBSMzqYJ4hhhMHZ/H01eO4/rllzHhludVsjIvC3imMyEtleG4qw/NSGZSTRIzTycFdbMWZkO7s+wSKd3+FnausZcNK+cjJHsE4YJ0xZgOAiLwInA/sSwTGmA/bPH8hMM3BeFRXRcdacwd9xsLxN1mrjcrXwua28wxvWM+NSYT8MdD3WI7pO57FM8bzbRV8s7WSFVur+GZrJa8uLeWZz60OX2y0lRyG53oSRF4qg3OSiY22MTmUl4TvsBBYCwJS++iEseo2JxNBHtC26E0pcMwhnn8N8HZHJ0RkOjAdoG/fvnbFp7pKxJp4zRoMo6+yjlVtgy2eoaRNn8EnfwLTSpQrmoGjpjHwlF9w4ShrzXtrq+HbihpWbK1kxdZKvtlayetflfHcF1aJ69goF0N6JTM8L5URntvgXkm+XZ9gjLV0dPj3bfrwQUpXDikbBMVksYhMA8YAJ3V03hgzC5gFMGbMmCCvaRBhUnrDsAutG0B9lTXPUDwPlj4F37wCJ/wUjv0JrpgeDMhKYkBWEucflQdYyWHzrlpPz8FKDm8tL+OFRVZyiIkSBuck7+s1DM9LZWivZOJjDpMc9u605jzCdX7AK7sQ1n9oFdeLCoKVWiokOZkItgJ92vyc7zl2ABE5HfgFcJIxpsHBeJQ/xKfAwNOs2zHXw3u/gg9+C0uehNPuhhGXHLDU0eUSCjITKchM5NwjcwEwxrBlVx3feBLDyrJK3lm5nRcXWx3MaJcwKCeZEXkp+5JDUe+UA5NDuJaWaC97GLQ2WRcKZofBnswqIJxMBIuBQSLSHysBTAYub/sEERkFPApMMMaEyU7pap/MgTD5OWuPhPl3wj+nwxePwFm/t1YeHYSI0Dcjgb4ZCZw9sjdgJYfS3XXWsFJZJd9sreK91TuZu8TarznKJQzKTmJYbirHDcjgvMbVxEBk9AjAmjDWRKB85FgiMMY0i8gNwHys5aOzjTErReQ3wBJjzOvAn4Ek4GWxliJuNsac51RMKkAKToBrP4Jv5sL7v4EnJ8LQc+CM33R6+0gRoU96An3SE5g4Yn9yKKusP2DO4eOSnby6rJSm+Pe5ICqRsvpkjkhx8LMFWuZgEJen1MRFgY5GhSgJiZrybYwZM8YsWbIk0GEoXzXWwucPwX/+19o7Ydy1VpE8m5Z4GmNYuGEX6a9cTF1NNRc0/objBmQwbXw/zijKcX7JaiA8epK1QuqUO63hOB/LhajwJiJLjTFjOjyniUAFRPUO+PB38OWzEJcCJ/0cxl5rLVm1w18GU9fvFGZnzuD5LzazdU8d2clxTB7bh8nj+pLbs4c97xMM9myBebdByTvQawSc8yDkjw50VCrIaCJQwWvHSlhwF6z/wLpa+fR7oPDc71y13CV1e+CP/ayhp+NvpqXV8FHxTuYs3MRHJW4EOHVoDtPG9+XEQVm4XN14r2BhjFU7at7PYe8OGDcdTr3LmrxXCk0EKhSsfc9KCO7VVo2js+6FPB9/q92yCJ44A6a8BEMmHHhqVy0vLNrM3CVbKN/bSJ/0Hlw+rh+XjskPfG0kO9RXwgf3wqLHILkXTPxT9xOrCguaCFRoaGm2hoo+/J21BeOIS60lpz37HP61bS17Bl6/EW760upldKCxuZX5K7czZ+Emvvh2F7FRLiYM78W08f0YW5CGhPoXZ+kSeONm2LECBk+ESX/u+p/jYVTVN1G2p44hOcmh/+cVATQRqNBSXwX/fcCaVAYY/2ProrTODnPM/wUsfhzuLAPX4a9KXrujmue+2Myry0qprm9mcE4SU4/px4VH55ESH8IXabU0wcJH4KM/AAKn/gLGXdetyeSahmbeW72DN5dv4+NiN40trRzTP52ZE4cyqm+afbEr22kiUKFpzxbrYrTlL1nF4065E0b94PBfZM9dYpW+uP4/XXq72sZm3vi6jDkLN/PN1koSYqM4/6hcph7Tj+F5qd34IAG2e5M1mbx2gVUc8NwHIe/oTr+8vqmFD9fs5M3l23h/zQ7qm1rJSYnj7BG55PaM5x8fb6B8bwMTh/fitrOGMCArzLYFDROaCFRo27oU5t9lFbvLGgpn3mtVSD3YcMQDIyB/LHx/ts9vubx0D3MWbuL1r8uob2rlyD49mXZMX849Mvfw5S2CkTGw6l/w9u3WsNu466weQlxyh09vaG7h05Jy3lxexrurdlDT2EJmUiwTh/fmnJG9GVuQvm+Svaahmcc//ZZZn6ynvrmVyWP7cPNpg8hOiffjB1SHo4lAhT5jYM2b8O7dsGsDHHEKnPW7/TX5vRpr4fe5Vu/hpJ93+20ra5t47ctS5izcxHp3DSnx0Xx/dB+mju8bmr/51ldaF/UtfsLaj2LSn2Ho2QA0tbTy2foK3vy6jHdWbqe6vpmeCTFMGNaLc4/M5Zj+6YfcV6J8bwN/f38tz32xmZgoFz/8Xn+mn3gEyaE8vBZGNBGo8NHcaI3/f/xHaKiCUdPglLusbTnB2j/h0RPhkqdh2AW2va33QrXnvtjE/JXbaWoxHDcgg6nH9OPMYSF4odqWxdZk8s6VVPQ5g8cSf8RLJa3srm0iOS6aM4f14pwje3PCwMwuf7ZNFTX8ZUEJb3xdRnpiLDeeOpDLj+nrWxVZZRtNBCr81O6CT/4Ci2ZBVCyccAsce4PVa3jtWvjxwv11eGzmrm5g7pIt+y5Uy/JcqHbZ2D7kpyU48p52am01LNu8m7e+2kza8se5tuUlWnHxTs4PST7xx5w4pJctw1/flFZy3zur+e+6Cvqk9+C2M4dw7sjc8LhuIwRpIlDhq2K9VeF09RuQnGvVLtr0Gfxiu31XKR9ES6vh45KdzFm4mQ+Ld2IMpCXE0DcjkX7pCfTLSKBvegL9MhLpl5FAdnJcwJZZGmNYXlrJG1+X8dY329hWWU9ctItTh2ZzyYBWTlr7B6I2vA+9j7Imk3OPsu19P11bzn1vr2HVtiqG5aYwc+JQvjcoy5b2VedpIlDhb9Nn1rLRsmWQMQhu9O+/kdLdtbz9zXa+rahhc0Utm3bVsHV3Ha1t/nvFx7jom55A33QrMbRNFHk9e9i7OxvWl/CqbVW8uXwbby3fxuZdtcRECScNzuKckbmcXpRDUly098mw8jV4eybUlls1i065E+LsmQdpbTW8sbyMP88vpnR3HScMzOT2CUMZkR/Cq7FCjCYCFRlaW2HNG9AjHfp/L9DR0NTSytbddWzaVcvmiho2VdR6HluJor6pdd9zXQK5PXt4koMnUaRbpbj7ZSTu/8LuhLU7qnlj+TbeXF7GBncNUS7h+IGZnDOyN2cV9SI14RCTt3V74P17YMlsSMn3TCZP6safwoEamlt4buFm/v7BWnbXNnHekbncduYQ+mYE/5BaqNNEoFSQMcbgrm5g065aNlV4EoX38a5adtU0HvD8jMRYa4+GdG+C2J8sspLj2FRRy5vLy3jj620U76hGBMb3z+CcI3szcXhv0hO7OEy2+Qt48xarvHXhuVapipRc2z5/VX0Tsz7ewOP/2UBLq2HqMf244dSBZIZDmY8gpYlAqRBTVd/EZk9SsJKDp0dRUcu2yu8OOXl7F2P6pXHOyN5MGtG7++v4W5rgs79bK7RcMXDaL2HsDzt1tXZn7aiq58H31/LS4i3ER7u47qQBXHNCfxK70AMKVq2thsaWVhpbWmlq9t5bx5paWmls9tzve2xo8pxr8Jzb97oWQ2NzK6cVZjMyv6dP8WgiUCqMNDa3Urp7/zDT5l219E6NZ9KI3s6U1961Ad76mVUhNvdoazK590hb32Ldzr38ZX4x76zcTmZSHDefPojJY/sEfFluY3MrGytqWLdzL+t27mXtzr1sr6yjsbmVRs8Xt/cL/YAv8BZDS6v93633XjCcaeP7+fRaTQRKqe4xBla8Cu/MtJbujr8eTr7Dtslkr2Wbd3PfvDUs2riL/pmJzDhrCBOH93J8tVVtYzMb3DWs3Vl9wJf+poraA77Q89N6kJ/Wg7joKGKiXMRFu4iJEmKiXMREu4iNchHrORYbFUVMtLQ55r2J53XWzXsuNsq17/ne47Gedq32XN36c9BEoJSyR91uePdXsOxpSO0Dp//auro7IcOapLdhdzRjDB+s2ckf31lDyY69HNmnJzMnDOXYARndbruytol1buvLfu2OvaxzW1/6pbvr9j0nyiX0y0hgUHYSA7OTGJSdzMDsJI7ISiQh9jCfzxhoboCmWmiuh6Y6z319B8fqPI/rrPPe+0O99sQZMNy3LUk1ESil7LXpc2sy2b3mwOPxqZCQaSWGfbd0SGx/zHOLTz1ozaiWVsNry0q5/90StlXWc/KQLG6fMJTC3oeuQmuMwb23Yd9v9m2/9N3VDfueFxft4oisJIZkxVOUDoNTWzkiuZnecU3ENFVZ5TjqK60r2L2P6yuhofoQX+b1gI/fqVFxEBMP0T0gxnOLjj/wfszVVp0tH2giUErZr7kRtnxhFbGrrej4VlNhXZfQ0thxG65oqyfhTQyJ300WjXFpvLG2gUeXVLKloQcTjzqCW88cTG5qD8oq61i3rYLSbTvYtn075RVu9uxyE9VYTYrUkkINmdH15PdoJCeukczoOlKlloTWGmKaqpGGKmjce5gPKlYJ9PhUiEu1CvXFxENMgucL2vvl3fZY+y/xhAOf1/750fHgcnY+RBOBUipwjIHGmoMni9oKqCm35h68P9ftAtPaYXN1JpbdJBNDC8nUEC9Nh357iULiU60v8vjU/V/q8akQ39PzBd/2WLvnxSY7/iXtD4dKBKG/RkspFdxErEnluCRI6+SKl9ZWqN/TYcJo2b0D96bNNEsMPZLTSUnLID0tk4TUjAO/yD1f7hKbqFt1HoYmAqVU8HG5rLmFhHRg0AGnkoAjAxJU+Ar9/o5SSqlu0USglFIRThOBUkpFOE0ESikV4TQRKKVUhNNEoJRSEU4TgVJKRThNBEopFeFCrsSEiLiBTT6+PBMotzGcUKCfOTLoZ44M3fnM/YwxWR2dCLlE0B0isuRgtTbClX7myKCfOTI49Zl1aEgppSKcJgKllIpwkZYIZgU6gADQzxwZ9DNHBkc+c0TNESillPquSOsRKKWUakcTgVJKRbiISQQiMkFEikVknYjMDHQ8ThORPiLyoYisEpGVInJzoGPyBxGJEpEvReTNQMfiDyLSU0ReEZE1IrJaRI4NdExOE5Gfev5NrxCRF0QkPtAx2U1EZovIThFZ0eZYuoi8KyJrPfdpdr1fRCQCEYkCHgImAkXAFBEpCmxUjmsGfmaMKQLGAz+JgM8McDOwOtBB+NGDwDvGmKFYG3eF9WcXkTzgJmCMMWY4EAVMDmxUjngKmNDu2EzgfWPMIOB9z8+2iIhEAIwD1hljNhhjGoEXgfMDHJOjjDHbjDHLPI+rsb4g8gIblbNEJB84G3g80LH4g4ikAicCTwAYYxqNMXsCGpR/RAM9RCQaSADKAhyP7YwxnwC72h0+H3ja8/hp4AK73i9SEkEesKXNz6WE+ZdiWyJSAIwCvghwKE57APg50BrgOPylP+AGnvQMhz0uIomBDspJxpitwF+AzcA2oNIYsyCwUflNjjFmm+fxdiDHroYjJRFELBFJAl4FbjHGVAU6HqeIyDnATmPM0kDH4kfRwNHAI8aYUUANNg4XBCPPuPj5WEkwF0gUkWmBjcr/jLXu37a1/5GSCLYCfdr8nO85FtZEJAYrCTxnjHkt0PE47HjgPBHZiDX0d6qIzAlsSI4rBUqNMd6e3itYiSGcnQ58a4xxG2OagNeA4wIck7/sEJHeAJ77nXY1HCmJYDEwSET6i0gs1uTS6wGOyVEiIlhjx6uNMfcHOh6nGWPuMMbkG2MKsP5+PzDGhPVvisaY7cAWERniOXQasCqAIfnDZmC8iCR4/o2fRphPkLfxOnCl5/GVwL/tajjaroaCmTGmWURuAOZjrTKYbYxZGeCwnHY8cAXwjYh85Tl2pzFmXuBCUg64EXjO8wvOBuDqAMfjKGPMFyLyCrAMa2Xcl4RhqQkReQE4GcgUkVLgV8B9wFwRuQarFP+ltr2flphQSqnIFilDQ0oppQ5CE4FSSkU4TQRKKRXhNBEopVSE00SglFIRThOBUu2ISIuIfNXmZtvVuiJS0LaipFLBICKuI1Cqi+qMMUcFOgil/EV7BEp1kohsFJE/icg3IrJIRAZ6jheIyAcislxE3heRvp7jOSLyTxH52nPzlkKIEpHHPDX1F4hIj4B9KKXQRKBUR3q0Gxq6rM25SmPMCOD/sKqdAvwdeNoYMxJ4Dvib5/jfgI+NMUdi1QDyXs0+CHjIGDMM2ANc7OinUeow9MpipdoRkb3GmKQOjm8ETjXGbPAU9NtujMkQkXKgtzGmyXN8mzEmU0TcQL4xpqFNGwXAu57NRRCR24EYY8y9fvhoSnVIewRKdY05yOOuaGjzuAWdq1MBpolAqa65rM39557Hn7F/u8SpwKeex+8D18O+vZRT/RWkUl2hv4ko9V092lRsBWtPYO8S0jQRWY71W/0Uz7EbsXYJm4G1Y5i3AujNwCxPtcgWrKSwDaWCjM4RKNVJnjmCMcaY8kDHopSddGhIKaUinPYIlFIqwmmPQCmlIpwmAqWUinCaCJRSKsJpIlBKqQiniUAppSLc/wOG9ov0FxhuPAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "7/7 [==============================] - 2s 159ms/step - loss: 0.0612 - binary_accuracy: 0.9787\n",
      "Metrics: 0.06116773933172226 0.9787499904632568\n"
     ]
    }
   ],
   "source": [
    "company_matches_model, loss, accuracy = train_find_matches(\n",
    "    x_train,\n",
    "    y_train,\n",
    "    x_test,\n",
    "    y_test,\n",
    "    x_val,\n",
    "    y_val,\n",
    "    learning_rate=0.0001,\n",
    "    epochs=11,\n",
    "    batch_size=100,\n",
    "    dropout=0.02,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In pattern recognition, information retrieval and classification (machine learning), precision (also called positive predictive value) is the fraction of relevant instances among the retrieved instances, while recall (also known as sensitivity) is the fraction of relevant instances that were retrieved."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 208,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "       False       0.96      0.97      0.97       996\n",
      "        True       0.97      0.96      0.97      1004\n",
      "\n",
      "    accuracy                           0.97      2000\n",
      "   macro avg       0.97      0.97      0.97      2000\n",
      "weighted avg       0.97      0.97      0.97      2000\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "error_analysis(\n",
    "    model=company_matches_model,\n",
    "    confidence=0.9,\n",
    "    x_test=x_test,\n",
    "    y_test=y_test,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Vectorizing: city#1\n",
      "Vectorizing: country#1\n",
      "Vectorizing: company_name#1\n",
      "Vectorizing: phone#1\n",
      "Vectorizing: street_address#1\n",
      "Vectorizing: website#1\n",
      "Vectorizing: linkedin_url#1\n",
      "Vectorizing: facebook_url#1\n",
      "Vectorizing: city#2\n",
      "Vectorizing: country#2\n",
      "Vectorizing: company_name#2\n",
      "Vectorizing: phone#2\n",
      "Vectorizing: street_address#2\n",
      "Vectorizing: website#2\n",
      "Vectorizing: linkedin_url#2\n",
      "Vectorizing: facebook_url#2\n",
      "Input: (3, 16, 96)\n",
      "Predictions: [[9.9999797e-01]\n",
      " [1.5853679e-13]\n",
      " [1.5527898e-09]]\n",
      "Confidence: [[ True]\n",
      " [False]\n",
      " [False]]\n"
     ]
    }
   ],
   "source": [
    "predictions: list = predict_find_matches(\n",
    "    model=company_matches_model, \n",
    "    records=[{\n",
    "        # Company 1\n",
    "        \"city#1\": \"San Francisco\",\n",
    "        \"country#1\": \"USA\",\n",
    "        \"company_name#1\": \"Airbnb\",\n",
    "        \"phone#1\": \"120-403-4032\",\n",
    "        \"street_address#1\": \"960 Air Ave.\",\n",
    "        \"website#1\": \"www.airbnb.com\",\n",
    "        \"linkedin_url#1\": \"https://www.linkedin.com/u/airbnb\",\n",
    "        \"facebook_url#1\": \"https://www.facebook.com/user/airbnb\",\n",
    "        # Company 2\n",
    "        \"city#2\": \"San Francisco\",\n",
    "        \"country#2\": \"USA\",\n",
    "        \"company_name#2\": \"Airbnb Inc.\",\n",
    "        \"phone#2\": \"120-403-4032\",\n",
    "        \"street_address#2\": \"960 Air Ave.\",\n",
    "        \"website#2\": \"www.airbnb.com\",\n",
    "        \"linkedin_url#2\": \"https://www.linkedin.com/u/airbnb\",\n",
    "        \"facebook_url#2\": \"https://www.facebook.com/user/airbnb\",\n",
    "    }, {\n",
    "        # Company 1\n",
    "        \"city#1\": \"San Francisco\",\n",
    "        \"country#1\": \"USA\",\n",
    "        \"company_name#1\": \"Airbnb\",\n",
    "        \"phone#1\": \"120-403-4032\",\n",
    "        \"street_address#1\": \"960 Air Ave.\",\n",
    "        \"website#1\": \"www.airbnb.com\",\n",
    "        \"linkedin_url#1\": \"https://www.linkedin.com/u/airbnb\",\n",
    "        \"facebook_url#1\": \"https://www.facebook.com/user/airbnb\",\n",
    "        # Company 2\n",
    "        \"city#2\": \"Mexico\",\n",
    "        \"country#2\": \"Mexico\",\n",
    "        \"company_name#2\": \"Taco Bell\",\n",
    "        \"phone#2\": \"543-543-1231\",\n",
    "        \"street_address#2\": \"50 Something St.\",\n",
    "        \"website#2\": \"www.taco-bell.com\",\n",
    "        \"linkedin_url#2\": \"https://www.linkedin.com/u/taco\",\n",
    "        \"facebook_url#2\": \"https://www.facebook.com/user/taco\",\n",
    "    }, {\n",
    "        # Company 1\n",
    "        \"city#1\": \"San Francisco\",\n",
    "        \"country#1\": \"USA\",\n",
    "        \"company_name#1\": \"Airbnb\",\n",
    "        \"phone#1\": \"120-403-4032\",\n",
    "        \"street_address#1\": \"960 Air Ave.\",\n",
    "        \"website#1\": \"www.airbnb.com\",\n",
    "        \"linkedin_url#1\": \"https://www.linkedin.com/u/airbnb\",\n",
    "        \"facebook_url#1\": \"https://www.facebook.com/user/airbnb\",\n",
    "        # Company 2\n",
    "        \"city#2\": \"Canada\",\n",
    "        \"country#2\": \"Canada\",\n",
    "        \"company_name#2\": \"McDonalds\",\n",
    "        \"phone#2\": \"432-435-6539\",\n",
    "        \"street_address#2\": \"43 Other Av.\",\n",
    "        \"website#2\": \"www.mcdonalds.com\",\n",
    "        \"linkedin_url#2\": \"https://www.linkedin.com/u/mcdonalds\",\n",
    "        \"facebook_url#2\": \"https://www.facebook.com/user/mcdonalds\",   \n",
    "    }],\n",
    "    confidence=0.99,\n",
    ")\n",
    "assert predictions[0][0]\n",
    "assert not predictions[1][0]\n",
    "assert not predictions[2][0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Persons Find Matches"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Persons: (10000, 15)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "raw_persons_dataframe: pd.DataFrame = generate_random_persons(\n",
    "    total=9900,\n",
    "    nulls=.01,\n",
    "    mutability=.01,\n",
    "    matches=.5,\n",
    "    relevance=50,\n",
    "    feedback=[\n",
    "        (\n",
    "            {\n",
    "                \"city\": \"San Francisco\",\n",
    "                \"country\": \"USA\",\n",
    "                \"name\": \"John Connor\",\n",
    "                \"phone\": \"120-403-4032\",\n",
    "                \"street_address\": \"960 Air Ave.\",\n",
    "                \"linkedin_url\": \"https://www.linkedin.com/u/john.connor\",\n",
    "                \"facebook_url\": \"https://www.facebook.com/user/john.connor\",\n",
    "            },\n",
    "            {\n",
    "                \"city\": \"San Francisco\",\n",
    "                \"country\": \"USA\",\n",
    "                \"name\": \"John Conor\",\n",
    "                \"phone\": \"120-403-4032\",\n",
    "                \"street_address\": \"960 Air Ave.\",\n",
    "                \"linkedin_url\": \"https://www.linkedin.com/u/john.connor\",\n",
    "                \"facebook_url\": \"https://www.facebook.com/user/john.connor\",\n",
    "            },\n",
    "            1\n",
    "        ),\n",
    "        (\n",
    "            {\n",
    "                \"city\": \"San Francisco\",\n",
    "                \"country\": \"USA\",\n",
    "                \"name\": \"John Connor\",\n",
    "                \"phone\": \"120-403-4032\",\n",
    "                \"street_address\": \"960 Air Ave.\",\n",
    "                \"linkedin_url\": \"https://www.linkedin.com/u/john.connor\",\n",
    "                \"facebook_url\": \"https://www.facebook.com/user/john.connor\",\n",
    "            },\n",
    "            {\n",
    "                \"city\": \"Los Angeles\",\n",
    "                \"country\": \"USA\",\n",
    "                \"name\": \"Sarah Connor\",\n",
    "                \"phone\": \"230-304-4322\",\n",
    "                \"street_address\": \"234 Water St.\",\n",
    "                \"linkedin_url\": \"https://www.linkedin.com/u/sarah.connor\",\n",
    "                \"facebook_url\": \"https://www.facebook.com/user/sarah.connor\",\n",
    "            },\n",
    "            0\n",
    "        ),\n",
    "    ]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>city#1</th>\n",
       "      <th>country#1</th>\n",
       "      <th>name#1</th>\n",
       "      <th>phone#1</th>\n",
       "      <th>street_address#1</th>\n",
       "      <th>linkedin_url#1</th>\n",
       "      <th>facebook_url#1</th>\n",
       "      <th>city#2</th>\n",
       "      <th>country#2</th>\n",
       "      <th>name#2</th>\n",
       "      <th>phone#2</th>\n",
       "      <th>street_address#2</th>\n",
       "      <th>linkedin_url#2</th>\n",
       "      <th>facebook_url#2</th>\n",
       "      <th>match</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td></td>\n",
       "      <td>Tier Crl</td>\n",
       "      <td>Lcenaueta Gseilmsr</td>\n",
       "      <td>815-637-3817</td>\n",
       "      <td>857 Loeoio Aouicl, Flizaw Av.</td>\n",
       "      <td>www.linkedin.com/u/ouenorra</td>\n",
       "      <td>https:///www.facebook.com/profile/ignsnmeerc</td>\n",
       "      <td>Gopalpuu</td>\n",
       "      <td></td>\n",
       "      <td>Usetnsnenosl Ipgogpclimm</td>\n",
       "      <td>814-431-3632</td>\n",
       "      <td>768 Gwamnpi Oeoe, Smsir Av.</td>\n",
       "      <td></td>\n",
       "      <td>https:///www.facebook.com/profile/iaergiearorouac</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Xboomtti</td>\n",
       "      <td>Udiispn</td>\n",
       "      <td>Oaaeunmeoa Ruueoccrr</td>\n",
       "      <td>630-744-5638</td>\n",
       "      <td>977 Gieiar Nilu, Rssutp St.</td>\n",
       "      <td>www.linkedin.com/u/bhrlmlonrucc</td>\n",
       "      <td>https:///www.facebook.com/profile/iujgopmosgr</td>\n",
       "      <td>Xboomtti</td>\n",
       "      <td>Udiispn</td>\n",
       "      <td>Oaaeunmeoa Ruueoccrr</td>\n",
       "      <td>630-744-5638</td>\n",
       "      <td>977 Gieiar Nilu, Rssutp St.</td>\n",
       "      <td>www.linkedis.com/u/bhrlmlonruccg</td>\n",
       "      <td>https:///www.facebook.com/profile/iujgopmosgr</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td></td>\n",
       "      <td>Annm Ipeu</td>\n",
       "      <td>Omrlgsuso Gmoensagaga</td>\n",
       "      <td>562-815-1012</td>\n",
       "      <td>25 Lpoemiou Ogsote, Aatigc St.</td>\n",
       "      <td>www.linkedin.com/u/potucoioioerakn</td>\n",
       "      <td>https:///www.facebook.com/profile/rootnrrea</td>\n",
       "      <td>Couli Ieo</td>\n",
       "      <td>Euoruknolop</td>\n",
       "      <td>Eaoejissuueo Algtcarl</td>\n",
       "      <td>267-867-6615</td>\n",
       "      <td>166 Lgieoi Gtrour, Trrumo Av.</td>\n",
       "      <td>www.linkedin.com/u/ltntscrnmsipat</td>\n",
       "      <td>https:///www.facebook.com/profile/roaitamiticioi</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Ggti Emaso</td>\n",
       "      <td>Tgeooagle</td>\n",
       "      <td>Iaaconmu Mriuiealsoum</td>\n",
       "      <td>785-109-8185</td>\n",
       "      <td>995 Oaceta Matfor, Rstr St.</td>\n",
       "      <td>www.linkedin.com/u/aeneormtougbiu</td>\n",
       "      <td>https:///www.facebook.com/profile/eealcpoc</td>\n",
       "      <td>Weldca</td>\n",
       "      <td>Opacioisoe</td>\n",
       "      <td>Avuwrripuk Slbaameai</td>\n",
       "      <td>474-219-1018</td>\n",
       "      <td>944 Isamlmmu Rnea, Nmofni Av.</td>\n",
       "      <td></td>\n",
       "      <td>https:///www.facebook.com/profile/imeeemcimo</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Oonioi Omadr</td>\n",
       "      <td>Uuucrarioy</td>\n",
       "      <td>Iccmelufce Ioiirneiuiec</td>\n",
       "      <td>511-342-3689</td>\n",
       "      <td>524 Acuoa Sogn, Amors Av.</td>\n",
       "      <td>www.linkedin.com/u/aoiihmigoeorona</td>\n",
       "      <td>https:///www.facebook.com/profile/aeroniun</td>\n",
       "      <td>Oonioi Omadr</td>\n",
       "      <td>Uuucrarioy</td>\n",
       "      <td>Iccmelufce Ioinrneiuiec</td>\n",
       "      <td>511-342-3689</td>\n",
       "      <td>524 Acuoa Sogn, Amors Av.</td>\n",
       "      <td>www.linkedin.com/u/aoiihmigoeorona</td>\n",
       "      <td>httls:o//www.facebook.com/profile/aeroniun</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Ipntgr Auteo</td>\n",
       "      <td>Uomscg Stsgit</td>\n",
       "      <td>Rgmuescguo Josgvcrcs</td>\n",
       "      <td>105-379-5222</td>\n",
       "      <td>876 Unair Ueaou, Ottpza Av.</td>\n",
       "      <td>www.linkedin.com/u/ofumigrlusar</td>\n",
       "      <td>https:///www.facebook.com/profile/hlitgsetmonii</td>\n",
       "      <td>Ipntgr Auteo</td>\n",
       "      <td>Uomscg Stsgit</td>\n",
       "      <td>Rgmuescguo Josgvcrcs</td>\n",
       "      <td>105-379-5222</td>\n",
       "      <td>876 Unair Ueaou, Ottpza Av.</td>\n",
       "      <td>www.linkedin.com/u/ofumigrlusar</td>\n",
       "      <td>https:///www.facebook.com/profile/hlitgsetmonii</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Urr Insatp</td>\n",
       "      <td>Roxn Ueo</td>\n",
       "      <td>Iclgbprpir Lromgonoeeae</td>\n",
       "      <td>833-460-2624</td>\n",
       "      <td>632 Sromg Uieor, Iila Av.</td>\n",
       "      <td>www.linkedin.com/u/iroaaeotlu</td>\n",
       "      <td>https:///www.facebook.com/profile/gtsuusnuenii</td>\n",
       "      <td>Eos Ovwc</td>\n",
       "      <td></td>\n",
       "      <td>Pgsnuera Erukuoetg</td>\n",
       "      <td>835-946-6096</td>\n",
       "      <td>646 Srslg Teuutm, Giosae Av.</td>\n",
       "      <td>www.linkedin.com/u/umrnlssoe</td>\n",
       "      <td>https:///www.facebook.com/profile/cungcctgpooitns</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Pin Aci</td>\n",
       "      <td>Rozil Aoraao</td>\n",
       "      <td>Ictioecrencu Comnlagt</td>\n",
       "      <td>386-628-3174</td>\n",
       "      <td>522 Tmqlda Uiacu, Glaom St.</td>\n",
       "      <td>www.linkedin.com/u/lnumlaiic</td>\n",
       "      <td>https:///www.facebook.com/profile/couuasre</td>\n",
       "      <td>Uucrt Mnmica</td>\n",
       "      <td>Rrsrproir</td>\n",
       "      <td>Eimuaitgar Egseisrpii</td>\n",
       "      <td>367-236-4533</td>\n",
       "      <td>193 Taimsetu Pieg, Lonrc St.</td>\n",
       "      <td>www.linkedin.com/u/sioriaro</td>\n",
       "      <td>https:///www.facebook.com/profile/eierrolatem</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Larai Rarant</td>\n",
       "      <td>Romr Lga</td>\n",
       "      <td>Lauenutpmars Csrtcngiorep</td>\n",
       "      <td>282-415-4632</td>\n",
       "      <td>784 Rpoplls Etrmcm, Ueoian Av.</td>\n",
       "      <td>www.linkedin.com/u/rgoaoeleu</td>\n",
       "      <td>https:///www.facebook.com/profile/egrcmrtaenepao</td>\n",
       "      <td>Emi Inimla</td>\n",
       "      <td></td>\n",
       "      <td>Lrlcsarcpm Amniergpmr</td>\n",
       "      <td>964-182-2215</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>https:///www.facebook.com/profile/gratpoaeg</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Gmaouaiuuor</td>\n",
       "      <td>Aso Unip</td>\n",
       "      <td>Uhacorclje Aogiiarotgls</td>\n",
       "      <td>765-434-4144</td>\n",
       "      <td>40 Noe Tiusue, Iomulu St.</td>\n",
       "      <td>www.linkedin.com/u/pliicinpptir</td>\n",
       "      <td>https:///www.facebook.com/profile/uullineitrgsme</td>\n",
       "      <td>Gmaouaiuuor</td>\n",
       "      <td>Aso Unip</td>\n",
       "      <td>Uhacorclja Aoginarotgls</td>\n",
       "      <td>765-434-4144</td>\n",
       "      <td>40 Noe Tiusue, Iomulu St.</td>\n",
       "      <td>www.linkedin.com/u/pliicinpptir</td>\n",
       "      <td>https:///www.facebook.com/profile/uullineitrgsle</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         city#1      country#1                     name#1       phone#1  \\\n",
       "0                     Tier Crl         Lcenaueta Gseilmsr  815-637-3817   \n",
       "1      Xboomtti        Udiispn       Oaaeunmeoa Ruueoccrr  630-744-5638   \n",
       "2                    Annm Ipeu      Omrlgsuso Gmoensagaga  562-815-1012   \n",
       "3    Ggti Emaso      Tgeooagle      Iaaconmu Mriuiealsoum  785-109-8185   \n",
       "4  Oonioi Omadr     Uuucrarioy    Iccmelufce Ioiirneiuiec  511-342-3689   \n",
       "5  Ipntgr Auteo  Uomscg Stsgit       Rgmuescguo Josgvcrcs  105-379-5222   \n",
       "6    Urr Insatp       Roxn Ueo    Iclgbprpir Lromgonoeeae  833-460-2624   \n",
       "7       Pin Aci   Rozil Aoraao      Ictioecrencu Comnlagt  386-628-3174   \n",
       "8  Larai Rarant       Romr Lga  Lauenutpmars Csrtcngiorep  282-415-4632   \n",
       "9   Gmaouaiuuor       Aso Unip    Uhacorclje Aogiiarotgls  765-434-4144   \n",
       "\n",
       "                 street_address#1                      linkedin_url#1  \\\n",
       "0   857 Loeoio Aouicl, Flizaw Av.         www.linkedin.com/u/ouenorra   \n",
       "1     977 Gieiar Nilu, Rssutp St.     www.linkedin.com/u/bhrlmlonrucc   \n",
       "2  25 Lpoemiou Ogsote, Aatigc St.  www.linkedin.com/u/potucoioioerakn   \n",
       "3     995 Oaceta Matfor, Rstr St.   www.linkedin.com/u/aeneormtougbiu   \n",
       "4       524 Acuoa Sogn, Amors Av.  www.linkedin.com/u/aoiihmigoeorona   \n",
       "5     876 Unair Ueaou, Ottpza Av.     www.linkedin.com/u/ofumigrlusar   \n",
       "6       632 Sromg Uieor, Iila Av.       www.linkedin.com/u/iroaaeotlu   \n",
       "7     522 Tmqlda Uiacu, Glaom St.        www.linkedin.com/u/lnumlaiic   \n",
       "8  784 Rpoplls Etrmcm, Ueoian Av.        www.linkedin.com/u/rgoaoeleu   \n",
       "9       40 Noe Tiusue, Iomulu St.     www.linkedin.com/u/pliicinpptir   \n",
       "\n",
       "                                     facebook_url#1        city#2  \\\n",
       "0      https:///www.facebook.com/profile/ignsnmeerc      Gopalpuu   \n",
       "1     https:///www.facebook.com/profile/iujgopmosgr      Xboomtti   \n",
       "2       https:///www.facebook.com/profile/rootnrrea     Couli Ieo   \n",
       "3        https:///www.facebook.com/profile/eealcpoc        Weldca   \n",
       "4        https:///www.facebook.com/profile/aeroniun  Oonioi Omadr   \n",
       "5   https:///www.facebook.com/profile/hlitgsetmonii  Ipntgr Auteo   \n",
       "6    https:///www.facebook.com/profile/gtsuusnuenii      Eos Ovwc   \n",
       "7        https:///www.facebook.com/profile/couuasre  Uucrt Mnmica   \n",
       "8  https:///www.facebook.com/profile/egrcmrtaenepao    Emi Inimla   \n",
       "9  https:///www.facebook.com/profile/uullineitrgsme   Gmaouaiuuor   \n",
       "\n",
       "       country#2                    name#2       phone#2  \\\n",
       "0                 Usetnsnenosl Ipgogpclimm  814-431-3632   \n",
       "1        Udiispn      Oaaeunmeoa Ruueoccrr  630-744-5638   \n",
       "2    Euoruknolop     Eaoejissuueo Algtcarl  267-867-6615   \n",
       "3     Opacioisoe      Avuwrripuk Slbaameai  474-219-1018   \n",
       "4     Uuucrarioy   Iccmelufce Ioinrneiuiec  511-342-3689   \n",
       "5  Uomscg Stsgit      Rgmuescguo Josgvcrcs  105-379-5222   \n",
       "6                       Pgsnuera Erukuoetg  835-946-6096   \n",
       "7      Rrsrproir     Eimuaitgar Egseisrpii  367-236-4533   \n",
       "8                    Lrlcsarcpm Amniergpmr  964-182-2215   \n",
       "9       Aso Unip   Uhacorclja Aoginarotgls  765-434-4144   \n",
       "\n",
       "                street_address#2                      linkedin_url#2  \\\n",
       "0    768 Gwamnpi Oeoe, Smsir Av.                                       \n",
       "1    977 Gieiar Nilu, Rssutp St.    www.linkedis.com/u/bhrlmlonruccg   \n",
       "2  166 Lgieoi Gtrour, Trrumo Av.   www.linkedin.com/u/ltntscrnmsipat   \n",
       "3  944 Isamlmmu Rnea, Nmofni Av.                                       \n",
       "4      524 Acuoa Sogn, Amors Av.  www.linkedin.com/u/aoiihmigoeorona   \n",
       "5    876 Unair Ueaou, Ottpza Av.     www.linkedin.com/u/ofumigrlusar   \n",
       "6   646 Srslg Teuutm, Giosae Av.        www.linkedin.com/u/umrnlssoe   \n",
       "7   193 Taimsetu Pieg, Lonrc St.         www.linkedin.com/u/sioriaro   \n",
       "8                                                                      \n",
       "9      40 Noe Tiusue, Iomulu St.     www.linkedin.com/u/pliicinpptir   \n",
       "\n",
       "                                      facebook_url#2  match  \n",
       "0  https:///www.facebook.com/profile/iaergiearorouac      0  \n",
       "1      https:///www.facebook.com/profile/iujgopmosgr      1  \n",
       "2   https:///www.facebook.com/profile/roaitamiticioi      0  \n",
       "3       https:///www.facebook.com/profile/imeeemcimo      0  \n",
       "4         httls:o//www.facebook.com/profile/aeroniun      1  \n",
       "5    https:///www.facebook.com/profile/hlitgsetmonii      1  \n",
       "6  https:///www.facebook.com/profile/cungcctgpooitns      0  \n",
       "7      https:///www.facebook.com/profile/eierrolatem      0  \n",
       "8        https:///www.facebook.com/profile/gratpoaeg      0  \n",
       "9   https:///www.facebook.com/profile/uullineitrgsle      1  "
      ]
     },
     "execution_count": 173,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "raw_persons_dataframe.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Vectorizing: city#1\n",
      "Vectorizing: country#1\n",
      "Vectorizing: name#1\n",
      "Vectorizing: phone#1\n",
      "Vectorizing: street_address#1\n",
      "Vectorizing: linkedin_url#1\n",
      "Vectorizing: facebook_url#1\n",
      "Vectorizing: city#2\n",
      "Vectorizing: country#2\n",
      "Vectorizing: name#2\n",
      "Vectorizing: phone#2\n",
      "Vectorizing: street_address#2\n",
      "Vectorizing: linkedin_url#2\n",
      "Vectorizing: facebook_url#2\n",
      "Vectorizing: match\n"
     ]
    }
   ],
   "source": [
    "vectorized_persons_dataframe: pd.DataFrame = df2vec(raw_persons_dataframe)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>city#1</th>\n",
       "      <th>country#1</th>\n",
       "      <th>name#1</th>\n",
       "      <th>phone#1</th>\n",
       "      <th>street_address#1</th>\n",
       "      <th>linkedin_url#1</th>\n",
       "      <th>facebook_url#1</th>\n",
       "      <th>city#2</th>\n",
       "      <th>country#2</th>\n",
       "      <th>name#2</th>\n",
       "      <th>phone#2</th>\n",
       "      <th>street_address#2</th>\n",
       "      <th>linkedin_url#2</th>\n",
       "      <th>facebook_url#2</th>\n",
       "      <th>match</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[-0.33806208, 0.086895615, 0.116224825, 0.0618...</td>\n",
       "      <td>[-0.34933227, -0.2106205, -0.1360507, 0.038428...</td>\n",
       "      <td>[0.16887774, 0.65122426, 0.19989617, -0.784912...</td>\n",
       "      <td>[-0.2289451, -0.44226894, 0.35295358, -0.10167...</td>\n",
       "      <td>[-0.71778774, -0.6098328, -0.13922524, -0.4531...</td>\n",
       "      <td>[0.48291373, 0.19668448, 0.58630955, -0.184213...</td>\n",
       "      <td>[-0.28951976, -0.5956406, -0.18228978, 0.19732...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[-0.06459761, -0.7431905, 0.33501643, 0.417832...</td>\n",
       "      <td>[0.10370531, 0.4567527, 0.1802727, -0.5701315,...</td>\n",
       "      <td>[-0.33848244, -0.3047452, 0.13612664, 0.013021...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[0.56222314, 0.2046036, 0.3803789, -0.24411993...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>[-0.80358577, -0.82671404, -0.37399483, 0.0263...</td>\n",
       "      <td>[0.1924375, -0.7932594, 0.5277333, 0.054435894...</td>\n",
       "      <td>[-0.15253712, -0.76584876, 0.40635836, 0.20158...</td>\n",
       "      <td>[0.06750767, 0.63540876, 0.24815121, -0.871682...</td>\n",
       "      <td>[-0.3043956, -0.15834653, 0.4382156, -0.308350...</td>\n",
       "      <td>[0.4694089, -0.11895573, 0.23903382, -0.576898...</td>\n",
       "      <td>[0.39928356, 0.3982561, 0.3712339, -0.10306541...</td>\n",
       "      <td>[-0.80358577, -0.82671404, -0.37399483, 0.0263...</td>\n",
       "      <td>[0.1924375, -0.7932594, 0.5277333, 0.054435894...</td>\n",
       "      <td>[-0.15253712, -0.76584876, 0.40635836, 0.20158...</td>\n",
       "      <td>[0.06750767, 0.63540876, 0.24815121, -0.871682...</td>\n",
       "      <td>[-0.3043956, -0.15834653, 0.4382156, -0.308350...</td>\n",
       "      <td>[0.5952095, -0.89793414, -0.43667608, -0.36790...</td>\n",
       "      <td>[0.39928356, 0.3982561, 0.3712339, -0.10306541...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[0.08995758, -0.44443446, 0.9511595, -0.394787...</td>\n",
       "      <td>[-0.13928975, -0.27118108, 0.17542964, 0.12549...</td>\n",
       "      <td>[0.17750491, 0.29854736, 0.18376015, -0.653003...</td>\n",
       "      <td>[-0.05129221, -0.254079, 0.23716325, -0.160038...</td>\n",
       "      <td>[0.1709955, -0.5495266, -0.15206802, -0.385299...</td>\n",
       "      <td>[0.17216352, 0.3157959, 0.34098616, -0.2978202...</td>\n",
       "      <td>[0.16284096, -0.5654431, 0.1832442, 0.05119581...</td>\n",
       "      <td>[-0.36701265, -0.6110436, -0.31378126, -0.1317...</td>\n",
       "      <td>[-0.21349409, -0.47541454, 0.4256311, 0.051735...</td>\n",
       "      <td>[0.42300138, 0.6515244, 0.30562407, -0.6020454...</td>\n",
       "      <td>[-0.1127262, -0.2523224, 0.4268349, -0.0037642...</td>\n",
       "      <td>[0.27468294, -0.42466375, 0.27251393, -0.02946...</td>\n",
       "      <td>[0.42491132, 0.378892, 0.17448303, -0.22951539...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>[-0.03747105, -0.64545083, 0.45971423, -0.1361...</td>\n",
       "      <td>[-0.019605681, -0.52111864, -0.82023907, -0.24...</td>\n",
       "      <td>[0.0845055, -0.41557404, 0.23800759, -0.101001...</td>\n",
       "      <td>[0.104056105, 0.7275071, 0.08199444, -0.666919...</td>\n",
       "      <td>[0.06518493, -0.62061125, 0.16085462, -0.28852...</td>\n",
       "      <td>[-0.3498528, -0.6881462, -0.28281814, 0.108114...</td>\n",
       "      <td>[0.4320628, 0.48924857, 0.13128908, -0.2300279...</td>\n",
       "      <td>[-0.46792638, -0.75368106, -1.0384717, -0.3422...</td>\n",
       "      <td>[-0.2539997, -0.29880625, 0.3714609, -0.202027...</td>\n",
       "      <td>[0.43388075, -0.3060286, 0.3854429, 0.02821834...</td>\n",
       "      <td>[0.35702285, 0.2852614, 0.35552317, -0.7096885...</td>\n",
       "      <td>[-0.16640346, -0.29747316, 0.58952004, 0.01931...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[0.30283895, 0.42740172, 0.29458892, -0.142781...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>[0.34219104, -0.22272833, 0.29944885, 0.224598...</td>\n",
       "      <td>[-0.3439262, -0.62242496, -0.5250815, -0.03490...</td>\n",
       "      <td>[-0.046257228, 0.28269833, -0.03186041, 0.3358...</td>\n",
       "      <td>[0.37935656, 0.5592426, 0.43632874, -0.7491644...</td>\n",
       "      <td>[-0.1519582, -0.25960264, 0.5826453, 0.0145043...</td>\n",
       "      <td>[-0.16140854, -0.4845662, -0.042230137, -0.713...</td>\n",
       "      <td>[0.37973335, 0.3609019, 0.14084454, -0.1895739...</td>\n",
       "      <td>[0.34219104, -0.22272833, 0.29944885, 0.224598...</td>\n",
       "      <td>[-0.3439262, -0.62242496, -0.5250815, -0.03490...</td>\n",
       "      <td>[-0.01973866, 0.15516709, -0.04195468, 0.02686...</td>\n",
       "      <td>[0.37935656, 0.5592426, 0.43632874, -0.7491644...</td>\n",
       "      <td>[-0.1519582, -0.25960264, 0.5826453, 0.0145043...</td>\n",
       "      <td>[-0.16140854, -0.4845662, -0.042230137, -0.713...</td>\n",
       "      <td>[0.021800702, 0.14864485, 0.5057089, -0.278883...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>[-0.04819388, -0.47787336, 0.07228744, 0.47974...</td>\n",
       "      <td>[0.3513934, -0.8199316, -0.15960664, 0.2498603...</td>\n",
       "      <td>[-0.071127236, -0.6470745, 0.36597723, 0.26762...</td>\n",
       "      <td>[0.23022214, 0.60848933, 0.37335533, -0.583143...</td>\n",
       "      <td>[-0.09096038, -0.27614173, 0.4532675, 0.112272...</td>\n",
       "      <td>[-0.18090202, -0.47888225, -0.45863667, 0.3942...</td>\n",
       "      <td>[0.3809796, 0.4046349, 0.45697004, -0.2910423,...</td>\n",
       "      <td>[-0.04819388, -0.47787336, 0.07228744, 0.47974...</td>\n",
       "      <td>[0.3513934, -0.8199316, -0.15960664, 0.2498603...</td>\n",
       "      <td>[-0.071127236, -0.6470745, 0.36597723, 0.26762...</td>\n",
       "      <td>[0.23022214, 0.60848933, 0.37335533, -0.583143...</td>\n",
       "      <td>[-0.09096038, -0.27614173, 0.4532675, 0.112272...</td>\n",
       "      <td>[-0.18090202, -0.47888225, -0.45863667, 0.3942...</td>\n",
       "      <td>[0.3809796, 0.4046349, 0.45697004, -0.2910423,...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>[0.028002739, -0.14396653, 0.39804342, 0.65591...</td>\n",
       "      <td>[-0.009212457, -0.3523032, -0.05493374, 0.2824...</td>\n",
       "      <td>[-0.39452845, -0.9032259, 0.38573176, -0.18034...</td>\n",
       "      <td>[0.17285466, 0.7129661, 0.07258873, -0.7009667...</td>\n",
       "      <td>[-0.16663668, -0.275417, 0.454531, -0.02130499...</td>\n",
       "      <td>[-0.49066398, -0.638654, -0.1672852, -0.116535...</td>\n",
       "      <td>[0.30249104, 0.30733243, 0.4634897, -0.252889,...</td>\n",
       "      <td>[-0.4624399, -0.48834938, 0.5233271, 0.1649792...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[-0.28741723, -0.4469352, -0.1348333, 0.337426...</td>\n",
       "      <td>[0.36982015, 0.57638836, 0.1743142, -0.4407454...</td>\n",
       "      <td>[-0.24428569, -0.19849406, 0.134287, 0.0457142...</td>\n",
       "      <td>[0.13282955, 0.32247537, 0.22941804, -0.307424...</td>\n",
       "      <td>[0.51179546, 0.19048485, 0.37220564, -0.134883...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>[0.119052105, -0.42437333, -0.021797732, 0.279...</td>\n",
       "      <td>[-0.26730558, -0.7814309, 0.40825295, 0.159558...</td>\n",
       "      <td>[-0.025333434, 0.053912833, 0.4973488, 0.07986...</td>\n",
       "      <td>[0.1512706, 0.6326774, 0.18761227, -0.6831428,...</td>\n",
       "      <td>[-0.08875673, -0.6058327, 0.04786034, -0.30467...</td>\n",
       "      <td>[-0.073726594, -0.28641355, 0.07312735, -0.111...</td>\n",
       "      <td>[0.47086936, 0.20701978, 0.42375764, -0.107518...</td>\n",
       "      <td>[0.36361396, -0.4375361, 0.108727805, -0.09042...</td>\n",
       "      <td>[-0.28385842, -0.838116, -0.028244376, 0.38156...</td>\n",
       "      <td>[-0.08614476, -0.281144, 0.1672039, -0.0550166...</td>\n",
       "      <td>[0.3070662, 0.43798333, 0.45019332, -0.5888532...</td>\n",
       "      <td>[-0.2231224, -0.3571708, 0.64790654, -0.307938...</td>\n",
       "      <td>[0.055589586, -0.49380398, -0.15615425, 0.2085...</td>\n",
       "      <td>[0.44999632, 0.44849038, 0.29552653, -0.124911...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>[-0.15670006, -0.6702647, 0.22256592, 0.092040...</td>\n",
       "      <td>[-0.018678248, -0.42819875, 0.38891488, 0.3165...</td>\n",
       "      <td>[-0.5739234, -0.37686154, -0.4005447, 0.152078...</td>\n",
       "      <td>[-0.015278975, 0.8310641, 0.15561733, -0.81360...</td>\n",
       "      <td>[-0.20096792, -0.4802725, 0.40363297, 0.014315...</td>\n",
       "      <td>[0.37458932, -0.071318015, 0.23487063, -0.1671...</td>\n",
       "      <td>[0.3589448, 0.27427676, 0.30102172, -0.2006835...</td>\n",
       "      <td>[0.13237628, -0.11484671, 0.38990182, 0.564695...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[-0.27279982, -0.35362482, -0.4143887, 0.40578...</td>\n",
       "      <td>[0.25900197, 0.76728386, 0.24819458, -0.654519...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[0.44428128, 0.35183963, 0.20430581, -0.368437...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>[-0.3329734, -0.8390591, -0.36070523, 0.109232...</td>\n",
       "      <td>[0.08097789, -0.7514024, 0.117207125, 0.037478...</td>\n",
       "      <td>[-0.30612764, -0.34588873, 0.23880264, 0.19313...</td>\n",
       "      <td>[0.16071129, 0.62369317, 0.19147961, -0.624332...</td>\n",
       "      <td>[0.030532757, -0.40217185, 0.44054654, -0.1805...</td>\n",
       "      <td>[0.33148295, -0.764873, -0.036764845, 0.034251...</td>\n",
       "      <td>[0.29456136, 0.36628827, 0.38634416, -0.391016...</td>\n",
       "      <td>[-0.3329734, -0.8390591, -0.36070523, 0.109232...</td>\n",
       "      <td>[0.08097789, -0.7514024, 0.117207125, 0.037478...</td>\n",
       "      <td>[0.35527104, -0.5354709, 0.5608023, -0.1471202...</td>\n",
       "      <td>[0.16071129, 0.62369317, 0.19147961, -0.624332...</td>\n",
       "      <td>[0.030532757, -0.40217185, 0.44054654, -0.1805...</td>\n",
       "      <td>[0.33148295, -0.764873, -0.036764845, 0.034251...</td>\n",
       "      <td>[0.5189635, 0.29307744, 0.36366206, -0.1238148...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              city#1  \\\n",
       "0  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "1  [-0.80358577, -0.82671404, -0.37399483, 0.0263...   \n",
       "2  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "3  [-0.03747105, -0.64545083, 0.45971423, -0.1361...   \n",
       "4  [0.34219104, -0.22272833, 0.29944885, 0.224598...   \n",
       "5  [-0.04819388, -0.47787336, 0.07228744, 0.47974...   \n",
       "6  [0.028002739, -0.14396653, 0.39804342, 0.65591...   \n",
       "7  [0.119052105, -0.42437333, -0.021797732, 0.279...   \n",
       "8  [-0.15670006, -0.6702647, 0.22256592, 0.092040...   \n",
       "9  [-0.3329734, -0.8390591, -0.36070523, 0.109232...   \n",
       "\n",
       "                                           country#1  \\\n",
       "0  [-0.33806208, 0.086895615, 0.116224825, 0.0618...   \n",
       "1  [0.1924375, -0.7932594, 0.5277333, 0.054435894...   \n",
       "2  [0.08995758, -0.44443446, 0.9511595, -0.394787...   \n",
       "3  [-0.019605681, -0.52111864, -0.82023907, -0.24...   \n",
       "4  [-0.3439262, -0.62242496, -0.5250815, -0.03490...   \n",
       "5  [0.3513934, -0.8199316, -0.15960664, 0.2498603...   \n",
       "6  [-0.009212457, -0.3523032, -0.05493374, 0.2824...   \n",
       "7  [-0.26730558, -0.7814309, 0.40825295, 0.159558...   \n",
       "8  [-0.018678248, -0.42819875, 0.38891488, 0.3165...   \n",
       "9  [0.08097789, -0.7514024, 0.117207125, 0.037478...   \n",
       "\n",
       "                                              name#1  \\\n",
       "0  [-0.34933227, -0.2106205, -0.1360507, 0.038428...   \n",
       "1  [-0.15253712, -0.76584876, 0.40635836, 0.20158...   \n",
       "2  [-0.13928975, -0.27118108, 0.17542964, 0.12549...   \n",
       "3  [0.0845055, -0.41557404, 0.23800759, -0.101001...   \n",
       "4  [-0.046257228, 0.28269833, -0.03186041, 0.3358...   \n",
       "5  [-0.071127236, -0.6470745, 0.36597723, 0.26762...   \n",
       "6  [-0.39452845, -0.9032259, 0.38573176, -0.18034...   \n",
       "7  [-0.025333434, 0.053912833, 0.4973488, 0.07986...   \n",
       "8  [-0.5739234, -0.37686154, -0.4005447, 0.152078...   \n",
       "9  [-0.30612764, -0.34588873, 0.23880264, 0.19313...   \n",
       "\n",
       "                                             phone#1  \\\n",
       "0  [0.16887774, 0.65122426, 0.19989617, -0.784912...   \n",
       "1  [0.06750767, 0.63540876, 0.24815121, -0.871682...   \n",
       "2  [0.17750491, 0.29854736, 0.18376015, -0.653003...   \n",
       "3  [0.104056105, 0.7275071, 0.08199444, -0.666919...   \n",
       "4  [0.37935656, 0.5592426, 0.43632874, -0.7491644...   \n",
       "5  [0.23022214, 0.60848933, 0.37335533, -0.583143...   \n",
       "6  [0.17285466, 0.7129661, 0.07258873, -0.7009667...   \n",
       "7  [0.1512706, 0.6326774, 0.18761227, -0.6831428,...   \n",
       "8  [-0.015278975, 0.8310641, 0.15561733, -0.81360...   \n",
       "9  [0.16071129, 0.62369317, 0.19147961, -0.624332...   \n",
       "\n",
       "                                    street_address#1  \\\n",
       "0  [-0.2289451, -0.44226894, 0.35295358, -0.10167...   \n",
       "1  [-0.3043956, -0.15834653, 0.4382156, -0.308350...   \n",
       "2  [-0.05129221, -0.254079, 0.23716325, -0.160038...   \n",
       "3  [0.06518493, -0.62061125, 0.16085462, -0.28852...   \n",
       "4  [-0.1519582, -0.25960264, 0.5826453, 0.0145043...   \n",
       "5  [-0.09096038, -0.27614173, 0.4532675, 0.112272...   \n",
       "6  [-0.16663668, -0.275417, 0.454531, -0.02130499...   \n",
       "7  [-0.08875673, -0.6058327, 0.04786034, -0.30467...   \n",
       "8  [-0.20096792, -0.4802725, 0.40363297, 0.014315...   \n",
       "9  [0.030532757, -0.40217185, 0.44054654, -0.1805...   \n",
       "\n",
       "                                      linkedin_url#1  \\\n",
       "0  [-0.71778774, -0.6098328, -0.13922524, -0.4531...   \n",
       "1  [0.4694089, -0.11895573, 0.23903382, -0.576898...   \n",
       "2  [0.1709955, -0.5495266, -0.15206802, -0.385299...   \n",
       "3  [-0.3498528, -0.6881462, -0.28281814, 0.108114...   \n",
       "4  [-0.16140854, -0.4845662, -0.042230137, -0.713...   \n",
       "5  [-0.18090202, -0.47888225, -0.45863667, 0.3942...   \n",
       "6  [-0.49066398, -0.638654, -0.1672852, -0.116535...   \n",
       "7  [-0.073726594, -0.28641355, 0.07312735, -0.111...   \n",
       "8  [0.37458932, -0.071318015, 0.23487063, -0.1671...   \n",
       "9  [0.33148295, -0.764873, -0.036764845, 0.034251...   \n",
       "\n",
       "                                      facebook_url#1  \\\n",
       "0  [0.48291373, 0.19668448, 0.58630955, -0.184213...   \n",
       "1  [0.39928356, 0.3982561, 0.3712339, -0.10306541...   \n",
       "2  [0.17216352, 0.3157959, 0.34098616, -0.2978202...   \n",
       "3  [0.4320628, 0.48924857, 0.13128908, -0.2300279...   \n",
       "4  [0.37973335, 0.3609019, 0.14084454, -0.1895739...   \n",
       "5  [0.3809796, 0.4046349, 0.45697004, -0.2910423,...   \n",
       "6  [0.30249104, 0.30733243, 0.4634897, -0.252889,...   \n",
       "7  [0.47086936, 0.20701978, 0.42375764, -0.107518...   \n",
       "8  [0.3589448, 0.27427676, 0.30102172, -0.2006835...   \n",
       "9  [0.29456136, 0.36628827, 0.38634416, -0.391016...   \n",
       "\n",
       "                                              city#2  \\\n",
       "0  [-0.28951976, -0.5956406, -0.18228978, 0.19732...   \n",
       "1  [-0.80358577, -0.82671404, -0.37399483, 0.0263...   \n",
       "2  [0.16284096, -0.5654431, 0.1832442, 0.05119581...   \n",
       "3  [-0.46792638, -0.75368106, -1.0384717, -0.3422...   \n",
       "4  [0.34219104, -0.22272833, 0.29944885, 0.224598...   \n",
       "5  [-0.04819388, -0.47787336, 0.07228744, 0.47974...   \n",
       "6  [-0.4624399, -0.48834938, 0.5233271, 0.1649792...   \n",
       "7  [0.36361396, -0.4375361, 0.108727805, -0.09042...   \n",
       "8  [0.13237628, -0.11484671, 0.38990182, 0.564695...   \n",
       "9  [-0.3329734, -0.8390591, -0.36070523, 0.109232...   \n",
       "\n",
       "                                           country#2  \\\n",
       "0  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "1  [0.1924375, -0.7932594, 0.5277333, 0.054435894...   \n",
       "2  [-0.36701265, -0.6110436, -0.31378126, -0.1317...   \n",
       "3  [-0.2539997, -0.29880625, 0.3714609, -0.202027...   \n",
       "4  [-0.3439262, -0.62242496, -0.5250815, -0.03490...   \n",
       "5  [0.3513934, -0.8199316, -0.15960664, 0.2498603...   \n",
       "6  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "7  [-0.28385842, -0.838116, -0.028244376, 0.38156...   \n",
       "8  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "9  [0.08097789, -0.7514024, 0.117207125, 0.037478...   \n",
       "\n",
       "                                              name#2  \\\n",
       "0  [-0.06459761, -0.7431905, 0.33501643, 0.417832...   \n",
       "1  [-0.15253712, -0.76584876, 0.40635836, 0.20158...   \n",
       "2  [-0.21349409, -0.47541454, 0.4256311, 0.051735...   \n",
       "3  [0.43388075, -0.3060286, 0.3854429, 0.02821834...   \n",
       "4  [-0.01973866, 0.15516709, -0.04195468, 0.02686...   \n",
       "5  [-0.071127236, -0.6470745, 0.36597723, 0.26762...   \n",
       "6  [-0.28741723, -0.4469352, -0.1348333, 0.337426...   \n",
       "7  [-0.08614476, -0.281144, 0.1672039, -0.0550166...   \n",
       "8  [-0.27279982, -0.35362482, -0.4143887, 0.40578...   \n",
       "9  [0.35527104, -0.5354709, 0.5608023, -0.1471202...   \n",
       "\n",
       "                                             phone#2  \\\n",
       "0  [0.10370531, 0.4567527, 0.1802727, -0.5701315,...   \n",
       "1  [0.06750767, 0.63540876, 0.24815121, -0.871682...   \n",
       "2  [0.42300138, 0.6515244, 0.30562407, -0.6020454...   \n",
       "3  [0.35702285, 0.2852614, 0.35552317, -0.7096885...   \n",
       "4  [0.37935656, 0.5592426, 0.43632874, -0.7491644...   \n",
       "5  [0.23022214, 0.60848933, 0.37335533, -0.583143...   \n",
       "6  [0.36982015, 0.57638836, 0.1743142, -0.4407454...   \n",
       "7  [0.3070662, 0.43798333, 0.45019332, -0.5888532...   \n",
       "8  [0.25900197, 0.76728386, 0.24819458, -0.654519...   \n",
       "9  [0.16071129, 0.62369317, 0.19147961, -0.624332...   \n",
       "\n",
       "                                    street_address#2  \\\n",
       "0  [-0.33848244, -0.3047452, 0.13612664, 0.013021...   \n",
       "1  [-0.3043956, -0.15834653, 0.4382156, -0.308350...   \n",
       "2  [-0.1127262, -0.2523224, 0.4268349, -0.0037642...   \n",
       "3  [-0.16640346, -0.29747316, 0.58952004, 0.01931...   \n",
       "4  [-0.1519582, -0.25960264, 0.5826453, 0.0145043...   \n",
       "5  [-0.09096038, -0.27614173, 0.4532675, 0.112272...   \n",
       "6  [-0.24428569, -0.19849406, 0.134287, 0.0457142...   \n",
       "7  [-0.2231224, -0.3571708, 0.64790654, -0.307938...   \n",
       "8  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "9  [0.030532757, -0.40217185, 0.44054654, -0.1805...   \n",
       "\n",
       "                                      linkedin_url#2  \\\n",
       "0  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "1  [0.5952095, -0.89793414, -0.43667608, -0.36790...   \n",
       "2  [0.27468294, -0.42466375, 0.27251393, -0.02946...   \n",
       "3  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "4  [-0.16140854, -0.4845662, -0.042230137, -0.713...   \n",
       "5  [-0.18090202, -0.47888225, -0.45863667, 0.3942...   \n",
       "6  [0.13282955, 0.32247537, 0.22941804, -0.307424...   \n",
       "7  [0.055589586, -0.49380398, -0.15615425, 0.2085...   \n",
       "8  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "9  [0.33148295, -0.764873, -0.036764845, 0.034251...   \n",
       "\n",
       "                                      facebook_url#2  match  \n",
       "0  [0.56222314, 0.2046036, 0.3803789, -0.24411993...      0  \n",
       "1  [0.39928356, 0.3982561, 0.3712339, -0.10306541...      1  \n",
       "2  [0.42491132, 0.378892, 0.17448303, -0.22951539...      0  \n",
       "3  [0.30283895, 0.42740172, 0.29458892, -0.142781...      0  \n",
       "4  [0.021800702, 0.14864485, 0.5057089, -0.278883...      1  \n",
       "5  [0.3809796, 0.4046349, 0.45697004, -0.2910423,...      1  \n",
       "6  [0.51179546, 0.19048485, 0.37220564, -0.134883...      0  \n",
       "7  [0.44999632, 0.44849038, 0.29552653, -0.124911...      0  \n",
       "8  [0.44428128, 0.35183963, 0.20430581, -0.368437...      0  \n",
       "9  [0.5189635, 0.29307744, 0.36366206, -0.1238148...      1  "
      ]
     },
     "execution_count": 175,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vectorized_persons_dataframe.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 209,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predictors: ['city#1', 'country#1', 'name#1', 'phone#1', 'street_address#1', 'linkedin_url#1', 'facebook_url#1', 'city#2', 'country#2', 'name#2', 'phone#2', 'street_address#2', 'linkedin_url#2', 'facebook_url#2']\n",
      "Predicted: match\n",
      "X (10000, 14)\n",
      "Y (10000,)\n",
      "Train (6400, 14) (6400,)\n",
      "Test (2000, 14) (2000,)\n",
      "Val (1600, 14) (1600,)\n",
      "Train: (6400, 14, 96) (6400,)\n",
      "Test: (2000, 14, 96) (2000,)\n",
      "Val: (1600, 14, 96) (1600,)\n",
      "Matches: [3202, 1001, 800] [6400, 2000, 1600]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "x_train, y_train, x_test, y_test, x_val, y_val = split_dataframe(\n",
    "    dataset=vectorized_persons_dataframe\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input: (6400, 14, 96)\n",
      "Model: \"model_11\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_12 (InputLayer)        [(None, 14, 96)]          0         \n",
      "_________________________________________________________________\n",
      "flatten_11 (Flatten)         (None, 1344)              0         \n",
      "_________________________________________________________________\n",
      "dense_55 (Dense)             (None, 4096)              5509120   \n",
      "_________________________________________________________________\n",
      "dropout_44 (Dropout)         (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dense_56 (Dense)             (None, 2048)              8390656   \n",
      "_________________________________________________________________\n",
      "dropout_45 (Dropout)         (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_57 (Dense)             (None, 1024)              2098176   \n",
      "_________________________________________________________________\n",
      "dropout_47 (Dropout)         (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_59 (Dense)             (None, 1)                 1025      \n",
      "=================================================================\n",
      "Total params: 15,998,977\n",
      "Trainable params: 15,998,977\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Epoch 1/11\n",
      "64/64 [==============================] - 36s 531ms/step - loss: 0.7799 - binary_accuracy: 0.5860 - val_loss: 0.5689 - val_binary_accuracy: 0.7330\n",
      "Epoch 2/11\n",
      "64/64 [==============================] - 32s 503ms/step - loss: 0.5327 - binary_accuracy: 0.7560 - val_loss: 0.3412 - val_binary_accuracy: 0.8810\n",
      "Epoch 3/11\n",
      "64/64 [==============================] - 31s 481ms/step - loss: 0.3660 - binary_accuracy: 0.8426 - val_loss: 0.2029 - val_binary_accuracy: 0.9270\n",
      "Epoch 4/11\n",
      "64/64 [==============================] - 22s 352ms/step - loss: 0.2357 - binary_accuracy: 0.9075 - val_loss: 0.1810 - val_binary_accuracy: 0.9170\n",
      "Epoch 5/11\n",
      "64/64 [==============================] - 20s 315ms/step - loss: 0.2133 - binary_accuracy: 0.9149 - val_loss: 0.6988 - val_binary_accuracy: 0.7675\n",
      "Epoch 6/11\n",
      "64/64 [==============================] - 17s 264ms/step - loss: 0.1926 - binary_accuracy: 0.9113 - val_loss: 0.1132 - val_binary_accuracy: 0.9530\n",
      "Epoch 7/11\n",
      "64/64 [==============================] - 19s 296ms/step - loss: 0.1434 - binary_accuracy: 0.9688 - val_loss: 0.1079 - val_binary_accuracy: 0.9600\n",
      "Epoch 8/11\n",
      "64/64 [==============================] - 18s 289ms/step - loss: 0.1165 - binary_accuracy: 0.9735 - val_loss: 0.3514 - val_binary_accuracy: 0.8850\n",
      "Epoch 9/11\n",
      "64/64 [==============================] - 18s 286ms/step - loss: 0.0477 - binary_accuracy: 0.9901 - val_loss: 0.1067 - val_binary_accuracy: 0.9545\n",
      "Epoch 10/11\n",
      "64/64 [==============================] - 19s 293ms/step - loss: 0.0529 - binary_accuracy: 0.9923 - val_loss: 0.1065 - val_binary_accuracy: 0.9560\n",
      "Epoch 11/11\n",
      "64/64 [==============================] - 19s 296ms/step - loss: 0.0602 - binary_accuracy: 0.9882 - val_loss: 0.1194 - val_binary_accuracy: 0.9530\n",
      "{'loss': [0.6908236742019653, 0.5013200640678406, 0.3346618711948395, 0.22607992589473724, 0.18611007928848267, 0.1386662721633911, 0.12322511523962021, 0.13855212926864624, 0.09261063486337662, 0.08545748889446259, 0.0826801136136055], 'binary_accuracy': [0.6195312738418579, 0.7653124928474426, 0.8578125238418579, 0.9123437404632568, 0.9290624856948853, 0.9426562786102295, 0.9704687595367432, 0.9646875262260437, 0.9864062666893005, 0.9868749976158142, 0.983593761920929], 'val_loss': [0.568852424621582, 0.34120988845825195, 0.20286764204502106, 0.18096677958965302, 0.6988037824630737, 0.11320533603429794, 0.10790549218654633, 0.3514396846294403, 0.10669606924057007, 0.10649818927049637, 0.11938215792179108], 'val_binary_accuracy': [0.7329999804496765, 0.8809999823570251, 0.9269999861717224, 0.9169999957084656, 0.7674999833106995, 0.953000009059906, 0.9599999785423279, 0.8849999904632568, 0.9545000195503235, 0.9559999704360962, 0.953000009059906]}\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "13/13 [==============================] - 2s 85ms/step - loss: 0.1154 - binary_accuracy: 0.9531\n",
      "Metrics: 0.11536917835474014 0.953125\n"
     ]
    }
   ],
   "source": [
    "persons_matches_model, loss, accuracy = train_find_matches(\n",
    "    x_train,\n",
    "    y_train,\n",
    "    x_test,\n",
    "    y_test,\n",
    "    x_val,\n",
    "    y_val,\n",
    "    learning_rate=0.0001,\n",
    "    epochs=11,\n",
    "    batch_size=100,\n",
    "    dropout=0.01,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 210,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "       False       0.96      0.97      0.97       999\n",
      "        True       0.97      0.96      0.97      1001\n",
      "\n",
      "    accuracy                           0.97      2000\n",
      "   macro avg       0.97      0.97      0.97      2000\n",
      "weighted avg       0.97      0.97      0.97      2000\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "error_analysis(\n",
    "    model=persons_matches_model,\n",
    "    confidence=0.9,\n",
    "    x_test=x_test,\n",
    "    y_test=y_test,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Vectorizing: city#1\n",
      "Vectorizing: country#1\n",
      "Vectorizing: name#1\n",
      "Vectorizing: phone#1\n",
      "Vectorizing: street_address#1\n",
      "Vectorizing: linkedin_url#1\n",
      "Vectorizing: facebook_url#1\n",
      "Vectorizing: city#2\n",
      "Vectorizing: country#2\n",
      "Vectorizing: name#2\n",
      "Vectorizing: phone#2\n",
      "Vectorizing: street_address#2\n",
      "Vectorizing: linkedin_url#2\n",
      "Vectorizing: facebook_url#2\n",
      "Input: (4, 14, 96)\n",
      "Predictions: [[9.9999774e-01]\n",
      " [8.2073718e-01]\n",
      " [9.9999374e-01]\n",
      " [1.4853030e-05]]\n",
      "Confidence: [[ True]\n",
      " [False]\n",
      " [ True]\n",
      " [False]]\n"
     ]
    }
   ],
   "source": [
    "predictions: list = predict_find_matches(\n",
    "    model=persons_matches_model, \n",
    "    records=[{\n",
    "        # Person 1\n",
    "        \"city#1\": \"San Francisco\",\n",
    "        \"country#1\": \"USA\",\n",
    "        \"name#1\": \"John Connor\",\n",
    "        \"phone#1\": \"120-403-4032\",\n",
    "        \"street_address#1\": \"960 Air Ave.\",\n",
    "        \"linkedin_url#1\": \"https://www.linkedin.com/u/john.connor\",\n",
    "        \"facebook_url#1\": \"https://www.facebook.com/user/john.connor\",\n",
    "        # Person 2\n",
    "        \"city#2\": \"San Francisco\",\n",
    "        \"country#2\": \"USA\",\n",
    "        \"name#2\": \"John Conor\",\n",
    "        \"phone#2\": \"120-403-4032\",\n",
    "        \"street_address#2\": \"960 Air Ave.\",\n",
    "        \"linkedin_url#2\": \"https://www.linkedin.com/u/john.connor\",\n",
    "        \"facebook_url#2\": \"https://www.facebook.com/user/john.connor\",\n",
    "    }, {\n",
    "        # Person 1\n",
    "        \"city#1\": \"San Francisco\",\n",
    "        \"country#1\": \"USA\",\n",
    "        \"name#1\": \"John Connor\",\n",
    "        \"phone#1\": \"120-403-4032\",\n",
    "        \"street_address#1\": \"960 Air Ave.\",\n",
    "        \"linkedin_url#1\": \"https://www.linkedin.com/u/john.connor\",\n",
    "        \"facebook_url#1\": \"https://www.facebook.com/user/john.connor\",\n",
    "        # Person 2\n",
    "        \"city#2\": \"Mexico\",\n",
    "        \"country#2\": \"Mexico\",\n",
    "        \"name#2\": \"Terminator\",\n",
    "        \"phone#2\": \"543-342-9594\",\n",
    "        \"street_address#2\": \"120 Terminator St.\",\n",
    "        \"linkedin_url#2\": \"https://www.linkedin.com/u/terminator\",\n",
    "        \"facebook_url#2\": \"https://www.facebook.com/user/terminator\",    \n",
    "    }, {\n",
    "        # Person 1\n",
    "        \"city#1\": \"Rosario\",\n",
    "        \"country#1\": \"Argentina\",\n",
    "        \"name#1\": \"Leo Celis\",\n",
    "        \"phone#1\": \"443-223-4423\",\n",
    "        \"street_address#1\": \"143 Nisman St.\",\n",
    "        \"linkedin_url#1\": \"https://www.linkedin.com/u/leo.celis\",\n",
    "        \"facebook_url#1\": \"https://www.facebook.com/user/leo.celis\",\n",
    "        # Person 2\n",
    "        \"city#2\": \"Rosario\",\n",
    "        \"country#2\": \"Argentina\",\n",
    "        \"name#2\": \"Leonardo Celis\",\n",
    "        \"phone#2\": \"543-342-9594\",\n",
    "        \"street_address#2\": \"142 Nisman Street\",\n",
    "        \"linkedin_url#2\": \"https://www.linkedin.com/u/leo.celis\",\n",
    "        \"facebook_url#2\": \"https://www.facebook.com/user/leo.celis\",\n",
    "    }, {\n",
    "        # Person 1\n",
    "        \"city#1\": \"San Francisco\",\n",
    "        \"country#1\": \"USA\",\n",
    "        \"name#1\": \"John Connor\",\n",
    "        \"phone#1\": \"120-403-4032\",\n",
    "        \"street_address#1\": \"960 Air Ave.\",\n",
    "        \"linkedin_url#1\": \"https://www.linkedin.com/u/john.connor\",\n",
    "        \"facebook_url#1\": \"https://www.facebook.com/user/john.connor\",\n",
    "        # Person 2\n",
    "        \"city#2\": \"Rosario\",\n",
    "        \"country#2\": \"Argentina\",\n",
    "        \"name#2\": \"Leonardo Celis\",\n",
    "        \"phone#2\": \"543-342-9594\",\n",
    "        \"street_address#2\": \"142 Nisman Street\",\n",
    "        \"linkedin_url#2\": \"https://www.linkedin.com/u/leo.celis\",\n",
    "        \"facebook_url#2\": \"https://www.facebook.com/user/leo.celis\",\n",
    "    }],\n",
    "    confidence=0.9,\n",
    ")\n",
    "assert predictions[0][0]\n",
    "assert not predictions[1][0]\n",
    "assert predictions[2][0]\n",
    "assert not predictions[3][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}